| Djannakhan | sub: blogged : http://blog.mansonthomas.com/2009/08/locale-configuration-issue-on-ubuntu.html ;) | 00:03 |
|---|---|---|
| Djannakhan | sub: thanks for the help ;) | 00:04 |
| Ng | are the official ec2 8.04 images available for download somewhere? | 00:19 |
| psi-jack_ | Sooo, anyone know of a Ubuntu-based distribution that is configured to be a Central User Management system? | 01:52 |
| psi-jack_ | eBox is working on it, but is not quiiite there yet. | 01:52 |
| qman__ | I actually downloaded ebox to try that | 01:59 |
| qman__ | haven't gotten into it yet, but even just looking at the site, it doesn't look good enough yet | 02:00 |
| psi-jack_ | They say it'll be a few weeks before they gear it up for that. | 02:00 |
| qman__ | quite frankly, that's one of the few things microsoft is still holding over the linux community--an easy, all-in-one directory service | 02:00 |
| psi-jack_ | I know, I was very specifically opening that can of worms. | 02:00 |
| psi-jack_ | pah! Easy? | 02:01 |
| psi-jack_ | Active Directory, EASY? Are you nuts? :) | 02:01 |
| qman__ | I guess that's an exaggeration, but it's arguably easier than setting it up the manual linux way | 02:01 |
| qman__ | but if something like ebox does come together | 02:01 |
| qman__ | that'll be a huge advantage | 02:01 |
| psi-jack_ | Eh.. yeah, you plop in the Windows 2008 R2 disc, and install | 02:01 |
| psi-jack_ | True. | 02:02 |
| psi-jack_ | So, | 02:02 |
| psi-jack_ | I think for now.. | 02:02 |
| psi-jack_ | i'm just going to use eBox and NIS+ for now. | 02:02 |
| qman__ | I've tried several times to get an openldap based setup working, but it's just such a big task | 02:03 |
| qman__ | automating the process would help a lot of people | 02:03 |
| clusty | psi-jack, nis+ ? | 02:13 |
| clusty | psi-jack, what package provides that? | 02:13 |
| clusty | psi-jack, still having trouble starting the damn nis server :( | 02:13 |
| twb | AFAIK NIS+ doesn't exist on GNU/Linux | 02:21 |
| twb | Also note that NIS+ is deprecated or obsolete even on Sun gear | 02:22 |
| twb | qman__: getting pam_ldap working on the client side is dead easy. It's the server side that badly needs turn-key level automation. | 02:22 |
| psi-jack | qman__: Yeah, it's sad too, cause, I used to know openldap like the back of my hand. | 02:33 |
| psi-jack | qman__: But that was back in 2.0 days. 2.4's a whooooole new ballgame. | 02:33 |
| qman__ | yeah | 02:59 |
| SJr | I have been looking for a guide for installing Ubuntu Server via USB, a guide I found is here: https://help.ubuntu.com/community/UbuntuServerFlashDriveInstaller, but I was wondering if there was actually a utility, I thought there was | 02:59 |
| qman__ | not to mention if you need an extra secure deployment, fully encrypted | 02:59 |
| qman__ | there's so many options, you have to first decide what to use | 02:59 |
| qman__ | kerberos, ssl | 02:59 |
| qman__ | and if you need to integrate with windows, you're really in trouble | 03:05 |
| twb | qman__: windows OR samba | 03:31 |
| SJr | How long should it take to create a LVM partition on a 1 TB drive? | 03:34 |
| qman__ | quite a while | 03:34 |
| qman__ | I can't say exactly, but that's a lot of space, should take anywhere from 20 minutes to an hour | 03:35 |
| qman__ | maybe longer | 03:35 |
| SJr | Why should I use LVM? | 03:38 |
| qman__ | LVM allows for greater flexibility in your partitioning scheme | 03:38 |
| qman__ | lets you treat multiple physical spaces as one unit, and create any number of partitions in it | 03:39 |
| SJr | Hmmmmmm | 03:39 |
| SJr | Encrypted home directories... | 03:40 |
| qman__ | for the sake of simplicity, I generally don't use LVM unless I need the features it enables | 03:40 |
| qman__ | but to each his own | 03:40 |
| qman__ | keep in mind formatting 1TB of space will take quite a while as well | 03:40 |
| SJr | hmmmm it seemed to have finished | 03:43 |
| SJr | which is excellent | 03:43 |
| SJr | because as god as my witness, judgement shall come to pass | 03:43 |
| twb | Only for sucky filesystems like ext3 | 03:43 |
| ball | twb: is ext4 equally sucky? | 03:43 |
| twb | In this capacity, yes, because it's backward-compatible with ext2 | 03:44 |
| qman__ | I use ext3 simply because it's never given me any trouble | 03:45 |
| qman__ | where XFS and reiser have | 03:45 |
| twb | qman__: I'm talking strictly about time to create the filesystem being O(n) not O(1) | 03:46 |
| qman__ | that's pretty subjective and they're certainly good filesystems, just my experience | 03:46 |
| twb | An operation being O(n) is not subjective! | 03:47 |
| qman__ | no, that's not what I was referring to | 03:47 |
| qman__ | I'm in an awkward position so my typing is slow | 03:47 |
| twb | OK, carry on then | 03:47 |
| qman__ | from the instances I've encountered, ext3 is incredibly good at handling corruption, reiser is ok but takes a lot more work, and XFS is a disaster | 03:49 |
| qman__ | I've lost entire XFS filesystems due to system crashes | 03:51 |
| oh_noes1 | Is there a package I can install to update Ubuntu Server 8.04 with validated Public Certificate authorities? | 03:51 |
| oh_noes1 | Im trying to "curl https://www.verisign.com" and it's not working | 03:51 |
| oh_noes1 | I assume because it doesnt ship with Verisign's (or any others) public CA | 03:52 |
| qman__ | yeah, I'm pretty sure there's a package with those in it | 03:52 |
| twb | XFS assumes you do not have crashes | 03:54 |
| oh_noes1 | pzl help me find it :s | 03:54 |
| twb | XFS is Wrong to use unless you have a UPS and write-thingies, which IIRC LVM doesn't support. | 03:54 |
| qman__ | I'm trying to find out what package it is | 03:54 |
| twb | Also XFS needs you to be running the latest upstream kernel version because otherwise your whole system will explode with bugs | 03:55 |
| qman__ | I have them installed on a machine | 03:55 |
| twb | But APART FROM THOSE MINOR THINGS, XFS is fantastic. So the XFS weenies tell me | 03:55 |
| qman__ | heh | 03:55 |
| twb | write barriers. Those are the things | 03:55 |
| qman__ | incidentally, I was running with a UPS | 03:55 |
| qman__ | but the system hard locked for some reason or another | 03:56 |
| twb | md RAID and/or LVM don't support write barriers. | 03:56 |
| qman__ | and poof, there goes my filesystem | 03:56 |
| twb | qman__: yup, btdt | 03:56 |
| qman__ | I simply cannot figure out which package this is | 03:56 |
| oh_noes1 | qman__: found it ... "ca-certificates" lol | 03:57 |
| oh_noes1 | so obivous | 03:57 |
| qman__ | I have a bunch of verisign and such certs in /etc/ssl/certs | 03:57 |
| qman__ | ah | 03:57 |
| qman__ | of course | 03:57 |
| twb | qman__: apt-file search .crt | 03:57 |
| twb | Or if it's already installed, dpkg -S or dlocate | 03:57 |
| qman__ | nice | 03:58 |
| qman__ | it's interesting you should mention that bit about LVM | 04:00 |
| qman__ | the tutorial on the ebox website instructs you to create an LVM and XFS filesystems | 04:00 |
| aubre | anyone have any luck running tsm on ubuntu server jaunty 64-bit? | 04:07 |
| uvirtbot | New bug: #415732 in bind9 (main) "Please sync bind9 1:9.6.1.dfsg.P1-3 (main) from Debian unstable (main)." [Wishlist,Confirmed] https://launchpad.net/bugs/415732 | 05:41 |
| cef | is it just me, or is security.ubuntu.com rather slow?? damn these kernel bugs. | 07:24 |
| twb | When oh when will we get our ubuntu/kSolaris ? | 07:26 |
| twb | >duck< | 07:26 |
| henkjan | cef: you can try to change security.ubuntu.com in /etc/apt/sources.list in yourcountrycode.ubuntu.com | 07:27 |
| cef | henkjan: yeah I could, and trust that the local mirror is either up to date (it lags), and that they have correct (valid, untampered) updates | 07:29 |
| henkjan | cef: ah. I did change sources.list to use the country mirror | 07:30 |
| henkjan | and I trust the country mirror, because we run it ourselfs :) | 07:30 |
| cef | hrm, 5% of the kernel before the connection to security.ubuntu terminates.. 3 times so far so 15% total.. guess we'll see | 07:31 |
| cef | henkjan: yeah I use the local mirror, but it has not 100% up to date for security | 07:31 |
| cef | amazing, this attempt, I'm almost getting 20KB/sec! | 07:35 |
| cef | spoke too soon.. down to 9k | 07:35 |
| cef | would be interesting to see how high the load figures are on security.ubuntu ;) | 07:42 |
| LiraNuna | what's up with http://security.ubuntu.com ? | 07:43 |
| LiraNuna | heh | 07:43 |
| twb | henkjan: I think that should be NN.archive.ubuntu.com | 07:45 |
| twb | Where NN is an ISO country code. | 07:45 |
| LiraNuna | those with 8.04 and non-zero can 'skip' the update? | 07:45 |
| LiraNuna | mmap > 0 | 07:46 |
| LiraNuna | I can't seem to find the linux-image-2.6.24-24-virtual package, | 07:51 |
| LiraNuna | ah, it's 32bit only | 07:55 |
| LiraNuna | figures... | 07:55 |
| uvirtbot | New bug: #254687 in vsftpd (main) "userlist options doesn't work in vsftpd" [Low,Confirmed] https://launchpad.net/bugs/254687 | 07:56 |
| artillerytx | Hey guys | 08:16 |
| artillerytx | Why is it i have to keep restarting apache2 to get my sub-domain to work ? | 08:18 |
| jtimberman | artillerytx: when you make changes to apache configuration, even for virtualhosts, you need to restart/reload apache. | 08:26 |
| artillerytx | jtimberman: i haven't made any changes since i last restarted it | 08:27 |
| jtimberman | artillerytx: i don't know what you mean then. | 08:27 |
| artillerytx | Okay it seems like my subdomain will work for a while then maybe a few hours later it stops working and the only way to get it working again is to restart apache2 | 08:28 |
| jtimberman | artillerytx: as in it becomes nonresponsive? Perhaps a traffic or performance issue? | 08:29 |
| artillerytx | it says server not found | 08:29 |
| _ruben | checking the logs would be a start | 08:30 |
| artillerytx | i don't see any errors in there | 08:31 |
| artillerytx | what am i looking for | 08:33 |
| \sh | moins | 08:34 |
| atomic__ | yeah, security.ubuntu.com seems hosed from here too | 08:38 |
| artillerytx | okay cool i tried to run apt-get update and it said couldn't connect to that domain | 08:39 |
| maswan | atomic__: yeah, we've been discussing it over in -mirrors too | 08:39 |
| atomic__ | btw, is it possible that local archives are not updated at the same time? im trying to update a server in a different country and apt-get ugprade seems to get the lists from the squid cache | 08:39 |
| atomic__ | thnx maswan | 08:39 |
| atomic__ | i mean apt-get update | 08:39 |
| maswan | but it requires the appropriate person at canonical to take actionn | 08:40 |
| atomic__ | its fairly early in the morning for the europe staff, they'll get to it :) | 08:40 |
| artillerytx | jtimberman: does http://swot.wwmcd.org work for you ? | 08:41 |
| jtimberman | not found by my DNS server. | 08:41 |
| artillerytx | so its a dns issue | 08:42 |
| artillerytx | so should i check my bind logs ? | 08:44 |
| cef | fwiw, I get like 5% of a file and then the connection to security dies. restart the download (using -d to apt-get) and it resumes for another 5% | 08:44 |
| atomic__ | things seem to have started moving | 09:04 |
| === scfh_ is now known as scfh | ||
| _ruben | stupid makefile variable expansion shit | 09:09 |
| _ruben | bah .. might have to have my makefile create a modified makefile and then call that one .. if only i could have immediate expansion of variables | 09:24 |
| acalvo | Hi | 09:33 |
| acalvo | I'm quite confused | 09:33 |
| acalvo | I have a server that acts as a PDC | 09:33 |
| acalvo | and it has a LDAP tree | 09:33 |
| acalvo | on the other side, I'm setting up a mail server | 09:33 |
| acalvo | I want to store mails in the mail server, under a fixed path | 09:34 |
| acalvo | I'm using dovecot and postfix | 09:34 |
| acalvo | but it seems that I'm misunderstanding several things | 09:35 |
| acalvo | I do not want to have virtual users | 09:35 |
| acalvo | but I don't know if the term "virtual users" applies to having the users in another server | 09:35 |
| acalvo | while the mail server is configured to lookup thru LDAP (to log in to the server using SSH, for instance) | 09:36 |
| Boohbah | yes that sounds very confusing | 09:42 |
| _ruben | yup .. confusing.. :) | 09:53 |
| _ruben | and virtual users is nearly always what you want .. unless all email accounts are local users, and all domains are to be treated identically | 09:54 |
| acalvo | well | 10:02 |
| acalvo | uf | 10:02 |
| acalvo | so confusing | 10:02 |
| _ruben | and well .. if you dont use dovecot's deliver lda but postfix' builtin local lda, mail by default ends up in /var/mail/<user> .. then again, that assumes you want all domains treated equally (if there'd be more than one) | 10:06 |
| _ruben | (<user> doesnt contain the domain, so user@domaina.com is the same as user@domainb.com | 10:07 |
| _ruben | ) | 10:07 |
| acalvo | well | 10:08 |
| acalvo | I do have more than one domain (domain1.maindomain.com, domain2.maindomain.com) | 10:08 |
| acalvo | but it does not matter | 10:09 |
| acalvo | the thing is that I'm not able to deliver mail | 10:09 |
| _ruben | show proof of that :) | 10:09 |
| _ruben | !doesnt work | 10:09 |
| ubottu | Doesn't work is a strong statement. Does it sit on the couch all day? Does it want more money? Is it on IRC all the time? Please be specific! Examples of what doesn't work tend to help too. | 10:09 |
| acalvo | Aug 19 11:10:40 jupiter postfix/smtpd[3222]: NOQUEUE: reject: RCPT from localhost[127.0.0.1]: 550 5.1.1 <andreas.calvo@admi.esci.es>: Recipient address rejected: User unknown in local recipient table; from=<andreas.calvo@admi.esci.es> to=<andreas.calvo@admi.esci.es> proto=SMTP helo=<andreas.calvo?admi.esci.es> | 10:11 |
| _ruben | so your recipient table isnt setup properly .. i have no experience with postfix talking to ldap though .. could you pastebin the output of postconf -n ? | 10:12 |
| acalvo | _ruben: http://pastebin.com/d782d9f02 | 10:13 |
| acalvo | but I'm using a documented guide in the ubuntu's community site | 10:13 |
| _ruben | (another hint is to start the setup using hash files instead of ldap, when that works, migrate those hash files to ldap one by one) | 10:14 |
| _ruben | i only see lists of aliases defined, no lists of valid mailboxes | 10:15 |
| acalvo | mmm let me see | 10:16 |
| acalvo | _ruben: well, it seems that I'm doing some progress | 10:20 |
| acalvo | now I see I'm login to the LDAP tree | 10:20 |
| acalvo | and it delivers to dovecot | 10:21 |
| acalvo | but it does not get the mail yet | 10:21 |
| _ruben | "does not get the mail yet" .. what does that mean? | 10:23 |
| acalvo | postfix can find the user | 10:23 |
| acalvo | and it relays to dovecot | 10:23 |
| acalvo | but still generates the non-delivery message | 10:24 |
| _ruben | pastebin the log lines that say so | 10:25 |
| acalvo | http://pastebin.com/pastebin.php | 10:28 |
| acalvo | sorry | 10:29 |
| acalvo | http://pastebin.com/d76682ac2 | 10:30 |
| _ruben | dovecot isnt accepting your email .. probably lacks a valid userlist as well | 10:31 |
| acalvo | mmm | 10:32 |
| acalvo | I can log in thru imap | 10:32 |
| acalvo | so it should be working | 10:32 |
| acalvo | what I don't understand is why is generating the non-delivery mail to acalvo@esci.es, which is uid@domain | 10:33 |
| _ruben | that's odd then .. dovecot isnt logging anything when postfix is trying to deliver mail to it ? | 10:33 |
| acalvo | _ruben: in a separate file, yes | 10:33 |
| acalvo | but it does not show anything relevant | 10:33 |
| acalvo | just says it's loading some plugins | 10:34 |
| _ruben | acalvo@esci.es is probably the return-path: used in the mail .. you didnt paste that part of the log ;) | 10:34 |
| _ruben | s/return-path/envelope sender/ | 10:35 |
| acalvo | one question: what does postfix needs to know when looking up the users? | 10:35 |
| acalvo | their mailbox directory? | 10:35 |
| acalvo | or what? | 10:35 |
| _ruben | nothing, just whether or not its valid | 10:35 |
| acalvo | mmm ok | 10:35 |
| _ruben | dovecot needs to know the details | 10:35 |
| _ruben | oh .. i see the problem now | 10:36 |
| _ruben | you have the domains listed in mydestination -> no virtual users | 10:37 |
| _ruben | nevermind | 10:37 |
| _ruben | local users are delivered via dovecot too | 10:37 |
| acalvo | aham | 10:38 |
| acalvo | so it should look them using dovecot? | 10:38 |
| acalvo | the thing is that I want to believe that postfix in this scenario is a dumb MTA, it just relays on dovecot | 10:38 |
| acalvo | so dovecot should do all the job | 10:38 |
| acalvo | but I think it's not | 10:38 |
| acalvo | postfix needs to know that the user is valid and then pass the mail to the relay agent | 10:38 |
| _ruben | relay=dovecot .. that looks odd to me .. it isnt treating it as local afterall | 10:39 |
| _ruben | i'd go for virtual domains instead personally .. using locals this way looks kinda nasty to me | 10:39 |
| _ruben | and you'll get some additional freedoms as well for "free" .. like being able to handle (certain) domains differently (might want/need it in the future :)) | 10:40 |
| acalvo | ok, seems fine | 10:40 |
| acalvo | since I do have some different domains | 10:40 |
| acalvo | any good howto to do that using LDAP? | 10:41 |
| _ruben | mailbox_command and mailbox_transport .. seems the latter overrides the first .. i assume you have a dovecot entry in your /etc/postfix/master.cf ? | 10:41 |
| acalvo | currently I was looking at https://help.ubuntu.com/community/Postfix/DovecotLDAP | 10:41 |
| acalvo | _ruben: yes | 10:41 |
| _ruben | the ldap part doesnt really matter .. any howto on it using mysql would be fine .. you'll just have to modify the lookup files to talk to ldap instead of mysql for instance | 10:42 |
| _ruben | but afaik, the problem is with dovecot, as postfix tries to hand it off to dovecot, but dovecot isnt accepting it | 10:42 |
| acalvo | at least now I'm seeing errors in dovecot's log file | 10:44 |
| _ruben | that's "good" :) | 10:54 |
| acalvo | now it's funny | 11:08 |
| acalvo | if I send a mail it is stored in mail_base/user (as in user@domain) | 11:08 |
| acalvo | but if I log in thru imap it looks in mail_base/uid | 11:09 |
| acalvo | so... it is saving mail but I've to tell dovecot to use the same pattern to look for mail | 11:09 |
| acalvo | look and save | 11:09 |
| \sh | guys, what is the best way to teach initramfs to deal with vlans? | 11:12 |
| soren | \sh: Why would the initramfs have to do that? | 11:13 |
| \sh | soren: it drops to a shell because it can't find an iscsi device, which is only reachable using my vlan network setup...I can see all interfaces != vlan interfaces in this shell | 11:14 |
| \sh | the real network setup is done after initramfs drops to the real root | 11:14 |
| \sh | which is wrong | 11:15 |
| \sh | in this case | 11:15 |
| soren | You've got root on iscsi? | 11:15 |
| \sh | soren: nope | 11:15 |
| soren | Why does this need to be in the initramfs then? | 11:16 |
| \sh | soren: no root...additional device somewhere mounted on /opt and it's written in /etc/fstab | 11:16 |
| soren | initramfs only mounts /. | 11:16 |
| \sh | soren: which is strange...it breaks before / and after fsck | 11:16 |
| soren | fsck? initramfs doesn't run fsck. | 11:17 |
| \sh | soren: so what breaks it then? | 11:17 |
| soren | It's hard to tell from here. | 11:17 |
| stefan___ | hello | 11:18 |
| soren | There is no fsck in initramfs, so if it's after an fsck, it's not initramfs. You're barking up the wrong tree. | 11:18 |
| soren | \sh: Your problem probably is that the network interface isn't being brought up early enough. | 11:19 |
| \sh | soren: ok then it's not initramfs but something doesn't like to bring up all network interfaces and that breaks iscsi to not find the devices, and the fsck.xfs wants to do its work, but fails and drops me into a root shell with only / mounted | 11:19 |
| soren | Ok. | 11:19 |
| soren | S40networking runs after iscsi. If your network does not get set up by udev, networking will not be available when iscsi starts. | 11:20 |
| soren | You can apply a hack to fix it, but the proper fix is quite difficult to get right.. | 11:21 |
| \sh | soren: so changing the bootorder of iscsi to start after networking should fix that | 11:21 |
| soren | You can start open-iscsi from an "up" clause in etc/network/interfaces of the interface in question, and then make sure the iscsi block device gets mounted at that point. | 11:22 |
| \sh | soren: that's the simple way...but looks like we should re-think the way of starting services which could need a full network config | 11:25 |
| soren | \sh: What we realy should do is fix the networking setup to be configured when udev discovers the physical devices in question. | 11:27 |
| \sh | soren: btw...is it possible to tell udev to setup bonds and vlan network interfaces somehow? | 11:27 |
| soren | \sh: Not exactly. | 11:30 |
| soren | udev triggers on availability of physical devices. | 11:30 |
| soren | vlan and bonding configuration needs to turned upside down, so to speak. | 11:31 |
| \sh | soren: ok...looks like that the problem lies somewhere else anyways../etc/init.d/open-iscsi is symlinked in /etc/network/if-up.d/ and afaik this will be called for all interfaces which will come up, right? | 11:32 |
| soren | Right now, it looks for "auto" clauses and finds e.g. "auto bond0", which referes to an bonded interface made up of two physical interfaces.. | 11:32 |
| soren | What we should do instead (or in addition) is to have the "iface eth0" stanza say that it's part of a bonded interface called "bond0". | 11:33 |
| soren | That way, bond0 would be configured as the relevant physical interfaces come up. | 11:33 |
| soren | \sh: Yes, that's correct. | 11:33 |
| \sh | soren: ok...that's the bugger...open-iscsi is not started during bootup, but will be executed for any network device which comes up | 11:34 |
| \sh | and it looks like that it breaks network setup when it tries to find the iscsi device on the if-uped interface, when open-iscsi can't reach the device | 11:37 |
| \sh | s/device/isci-device/ | 11:37 |
| soren | It shouldn't. | 11:39 |
| soren | It's meant to try to connect when each interface comes up and handle it all gracefully. | 11:40 |
| soren | Anyhow, I need lunch. | 11:40 |
| \sh | oh my...now it's getting really hard...rcS.d/S25open-iscsi | 11:40 |
| _ruben | acalvo: was out for lunch .. are you using prefetching in dovecot (one query for both user and passwd data)? if so, you're passwd query might give back the wrong locating, whereas the user query does not | 11:49 |
| acalvo | _ruben: well, I've enabled it a while back | 11:50 |
| acalvo | well, I've find a workaround | 11:53 |
| acalvo | using the %n variable didn't work | 11:53 |
| acalvo | because in one scenario it referred to the user part in user@domain | 11:54 |
| acalvo | and in the other referred to the uid of the user | 11:54 |
| _ruben | nice | 11:54 |
| acalvo | now I've try to use %i, which is the id of the user under unix | 11:54 |
| acalvo | is really nasty | 11:54 |
| acalvo | but it's working | 11:54 |
| maswan | \sh: oh btw, do you have p410i controllers in your blades? | 11:55 |
| \sh | maswan: nope e200i | 11:56 |
| \sh | maswan: we don't need much io performance on the blades itself :) | 11:57 |
| \sh | drive io performance ;) | 11:57 |
| maswan | \sh: Ok, I guess I'm the only one with those then too. :) | 11:59 |
| maswan | \sh: That's what's onboard, just like the 10GE nics. | 11:59 |
| \sh | maswan: in our dl365er we are using p400i dunno what's the difference between p400i and p410i | 12:00 |
| maswan | \sh: yeah, we have some of those. these are the first one with these, and I was just wondering if you could reproduce a cciss issue for me. :) | 12:01 |
| \sh | maswan: sorry..I would like to..but can't...anyways what's happening? | 12:06 |
| maswan | \sh: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/413070 | 12:12 |
| uvirtbot | Launchpad bug 413070 in linux "karmic cciss: error messages on working device" [Undecided,New] | 12:12 |
| \sh | maswan: ok...what I can tell you, on the e200i it doesn't log any messages like that | 12:20 |
| \sh | maswan: and on my dl365 with a p400i also no sign of those | 12:21 |
| maswan | \sh: I have p400is too, noproblem there | 12:24 |
| * \sh needs to find a good fix for this open-iscsi problem | 12:26 | |
| acalvo | _ruben: noew everything works | 12:30 |
| acalvo | just one question | 12:30 |
| acalvo | I need to have several subdomains | 12:30 |
| acalvo | but I only can log in using the main address | 12:30 |
| acalvo | one user can have more than one mail address | 12:30 |
| acalvo | so user@domain.com works | 12:31 |
| acalvo | but user@sub.domain.com doesn't | 12:31 |
| acalvo | althought it should look it up in the ldap tree | 12:31 |
| _ruben | acalvo: are those aliases, seperate mailboxes, something else? | 12:33 |
| acalvo | no | 12:33 |
| acalvo | I want to use the same mailbox for the every user | 12:33 |
| acalvo | althought a user could have more than one mail address | 12:33 |
| acalvo | it's like an alias | 12:33 |
| acalvo | but it's stored under the mail attribute of the ldap tree | 12:33 |
| _ruben | a mailbox is associated to a username, and can have multiple email addresses bound to it .. to login to the mailbox, you use the username, not email address(es) | 12:34 |
| _ruben | then again, it all depends on how you have stuff configured :) | 12:34 |
| acalvo | that last sentence didn't sound so good :/ | 12:34 |
| _ruben | well .. in my case, usernames are (primary) email adresses .. ppl would login using their primary email address, yet receive mail for any aliases as well (in that same mailbox) | 12:36 |
| acalvo | that's what I want | 12:36 |
| _ruben | how are your mailboxes "defined" ? based on username or username@domain? | 12:37 |
| RoyK | acalvo: what mail system_ | 12:37 |
| RoyK | ? | 12:37 |
| acalvo | _ruben: username@domain | 12:37 |
| acalvo | RoyK: dovecot+postfix with ldap backend | 12:38 |
| _ruben | acalvo: im guessing that in your current setup, you dont really have "aliases", but each "alias" has become its own mailbox | 12:38 |
| RoyK | acalvo: there is a #dovecot channel if you have problems with that | 12:38 |
| acalvo | RoyK: I know, and I've been there | 12:38 |
| _ruben | you'll need a way to differentiate between an alias and a mailbox | 12:38 |
| _ruben | which depends on your ldap schema | 12:39 |
| acalvo | _ruben: the thing is, the system is set up to deliver the mail not based on the address of the user, but the unix id | 12:39 |
| acalvo | so if a user has more than one address | 12:39 |
| _ruben | you can then have postfix send mail for any aliases to the apropriate mailbox | 12:39 |
| acalvo | it will be redirected to the user id | 12:39 |
| _ruben | acalvo: why would you want users to be able to login using an alias? | 12:52 |
| acalvo | well | 12:52 |
| acalvo | I'm working for a university | 12:52 |
| acalvo | and some people need to have more than one addres | 12:52 |
| acalvo | because they're in more than one department | 12:53 |
| acalvo | for example | 12:53 |
| _ruben | one mailbox ... multiple aliases ... only one login required | 12:53 |
| _ruben | thats the whole idea of aliases | 12:53 |
| acalvo | yes, I know | 12:53 |
| acalvo | and, in the old server (which I configured like 3 years ago), it worked | 12:54 |
| acalvo | not nice | 12:54 |
| acalvo | but it worked | 12:54 |
| acalvo | looking up every mail attribute for a user in ldap | 12:54 |
| acalvo | now, trying to use new ways to do that better | 12:54 |
| acalvo | it only works with the main domain address | 12:54 |
| acalvo | while the user has more than one address | 12:54 |
| acalvo | so user@domain.com works | 12:55 |
| acalvo | but the same user, which has user@sub.domain.com, does not work | 12:55 |
| acalvo | and it's quite annoying | 12:55 |
| _ruben | wait .. define "works" .. are we talking sending mail (smtp/postfix), or receiving mail (pop3|imap/dovecot)? | 12:55 |
| acalvo | sending mail | 12:55 |
| _ruben | ahh | 12:56 |
| acalvo | I can see how postfix tries to find the user | 12:56 |
| acalvo | but it always fails | 12:56 |
| _ruben | show logs? :) | 12:56 |
| acalvo | pastebin thinks I'm posting spam | 12:57 |
| _ruben | try http://pastebin.ubuntu.com or so :) | 12:57 |
| acalvo | http://pastebin.com/d303e802d | 12:58 |
| acalvo | replaced the @ with -AT- | 12:58 |
| _ruben | my guess would be flawed ldap lookup queries or something similar .. i find postfix' debug logs rather unreadable .. in my case (mysql based backend) i'd either sniff the mysql traffic or enable query logging for mysql .. not sure if ldap provides similar mechanisms | 13:01 |
| _ruben | bbiab | 13:01 |
| acalvo | I know | 13:02 |
| acalvo | funny thing is that I'm receiving twice the mails | 13:02 |
| acalvo | if I send them to the internal mailboxs | 13:02 |
| acalvo | any good webmail? | 13:41 |
| acalvo | Now we're using horde, but we must upgrade | 13:41 |
| acalvo | so we can change it for something newer | 13:42 |
| _ruben | im still "shopping" for a decent webmail interface .. roundcube looks nice, squirrelmail is the classic ofcourse, there's atmail too | 13:55 |
| sgsax | acalvo: "good" is very subjective | 13:59 |
| sgsax | I'd say most webmail apps are "good enough" | 13:59 |
| sgsax | they all kinda suck in one way or another, including gmail | 13:59 |
| _ruben | nobody's perfect ;) | 13:59 |
| sgsax | gmail fanbois would disagree with you :) | 14:00 |
| _ruben | hehe | 14:00 |
| sgsax | the uni I work at used a modified horde for a long time, had some serious scaling problems | 14:01 |
| sgsax | now we use hosted zimbra, which is interesting, to say the least | 14:01 |
| _ruben | horde didnt appeal much to me last i checked | 14:02 |
| sgsax | like I said, it's about as good as any of the rest of them | 14:02 |
| sgsax | my hosting provider lets me choose between horde, roundcube, and squirrelmail | 14:03 |
| _ruben | lets just use OWA ;) | 14:03 |
| sgsax | there's good and bad about each of them | 14:03 |
| _ruben | i'll probably offer something similar (the choice of a few) | 14:03 |
| sgsax | if you want simple, squirrelmail is the way to go | 14:04 |
| sgsax | it's about as basic as you can get | 14:04 |
| _ruben | fuck .. got a 1G / (on a 2G flash disk) .. cant do-release-upgrade from intrepid to jaunty | 14:19 |
| pmatulis | language please | 14:30 |
| * Boohbah gives pmatulis language | 14:35 | |
| sgsax | para espanol, pulse 1 | 14:37 |
| sgsax | pour le franncais, appuyez sur 2 | 14:37 |
| sgsax | voor Nederlands, druk op 3 | 14:38 |
| _ruben | hrm .. /boot refuses to get mounted during boot .. both when using uuid and label in fstab | 14:42 |
| sub | pmatulis: apt-get install language-pack-en? :P | 14:43 |
| sub | _ruben: What does mount say when you try to mount it? | 14:43 |
| _ruben | mount: special device /dev/disk/by-label/sys-boot does not exist | 14:45 |
| _ruben | its uuid isnt present under by-uuid either | 14:45 |
| sub | have you checked dmesg for any pertinent information? | 14:48 |
| _ruben | hmm .. blkid shows a different uuid than vol_id | 14:51 |
| _ruben | dmesg doesnt show anything obvious | 14:52 |
| sub | have you tried mounting it using the partition's device? | 14:54 |
| smoser | _ruben, so you see that disk/partition somewhere ? | 14:55 |
| smoser | ie, is it in /proc/partitions ? | 14:55 |
| Boohbah | sgsax: i've seen roundcube exploited in the wild quite a bit recently | 14:55 |
| slestak | SEENNICK cjwatson | 14:56 |
| _ruben | smoser: yes, it lists sda1 which is /boot .. its not under /dev/disk/by-{label,uuid} , but it is under /dev/disk/by-id and by-path | 14:57 |
| jetsaredim | is there a package for temperature/sensor monitoring for cli? | 14:57 |
| smoser | jetsaredim, lm-sensors ? | 14:58 |
| sgsax | Boohbah: I don't manage it, my hosting provider provides it | 14:59 |
| sgsax | and I rarely use it | 14:59 |
| smoser | _ruben, can you mount it by device ? (not as a permenant solution, just wondering) | 14:59 |
| _ruben | yup .. using /dev/sda1 i can mount it just fine | 15:00 |
| sgsax | _ruben: bah, then who needs HAL anyway? | 15:00 |
| _ruben | pata flashdrive .. onboard sata controllers .. addon raidcard .. dont trust sdXY for my partitions in such cases :) | 15:01 |
| RoyK | jetsaredim: type 'sensors' | 15:02 |
| RoyK | jetsaredim: you need to setup lmsensors first, though | 15:02 |
| jetsaredim | it seems right | 15:02 |
| jetsaredim | temperature sensor reporting 30 deg | 15:03 |
| RoyK | some of those sensor drivers aren't very good, so try to find the right one for your hardware | 15:04 |
| RoyK | but if it works, don't fix it | 15:04 |
| jetsaredim | yea | 15:04 |
| smoser | _ruben, is it ext[23] filesystem ? does e2label /dev/sda1 report a label? | 15:04 |
| Sam-I-Am | anyone here good with building packages? | 15:05 |
| Sam-I-Am | trying to figure something out... | 15:05 |
| maxb | #ubuntu-motu is packaging help central | 15:05 |
| Sam-I-Am | mkay thx | 15:05 |
| _ruben | smoser: ext2, and yes it does report a label | 15:06 |
| smoser | hm... then i'm of no more help i think. i just verified that setting a label and then removing it populates /dev/disk/by-label and then removes the entry | 15:07 |
| smoser | ie, here that "just works". | 15:07 |
| smoser | any messages in dmesg about sda ? | 15:07 |
| _ruben | nothing spectacular: http://pastebin.ubuntu.com/255754/ | 15:08 |
| smoser | maybe udev is hosed... | 15:09 |
| smoser | sudo udevadm trigger --verbose --dry-run --subsystem-match=block | 15:10 |
| smoser | does that show sda1 ? and maybe try without 'dry-run' to see if that does anything for you | 15:10 |
| _ruben | /sys/devices/pci0000:00/0000:00:07.1/host4/target4:0:0/4:0:0:0/block/sda/sda1 | 15:10 |
| _ruben | no change under /dev/disk/ | 15:11 |
| smoser | outside of that, i think i'm out of ideas... if udev is up, it "should" be populating that stuff for you. | 15:11 |
| smoser | you ran without '--dry-run' ? | 15:11 |
| _ruben | yes | 15:11 |
| smoser | hm... | 15:11 |
| smoser | yeah, i'm out of ideas. sorry | 15:11 |
| _ruben | thanks for trying anyway :) | 15:13 |
| smoser | one more thing i guess... now i'm just confused | 15:14 |
| smoser | sudo blkid | 15:14 |
| _ruben | it lists it by uuid and label | 15:14 |
| _ruben | yet vol_id lists another uuid | 15:14 |
| _ruben | hmm .. vol_id isnt working now | 15:15 |
| smoser | what release? | 15:20 |
| smoser | it kind of smells like vol_id is borked, which udev would rely on (i think). | 15:20 |
| smoser | https://bugs.launchpad.net/ubuntu/+source/udev/+bug/337015 | 15:20 |
| uvirtbot | Launchpad bug 337015 in udev "vol_id uuid detection regression" [Low,Fix released] | 15:20 |
| Psi-Jack | This is odd.. | 15:20 |
| Psi-Jack | I created my ssh keys from my desktop computer, which is kubuntu-9.04, transferred the keys to my router, added a password there, and then transferred that copy over to the rest of my servers. | 15:21 |
| Psi-Jack | When I ssh from my router to my desktop, it asks for the password for the private key, which is expected. When I ssh from desktop to router, it's passwordless, also expected. But when I ssh to any other system in my network, it goes keyboard interactive, not even using the keys. | 15:22 |
| sgsax | Psi-Jack: check value of PubkeyAuthentication in /etc/ssh/sshd_config | 15:23 |
| Psi-Jack | It's yes | 15:24 |
| sgsax | also, different versions of sshd will have a different value for AuthorizedKeysFile | 15:24 |
| sgsax | so if there's any question, uncomment that line in the same file and specify the path | 15:24 |
| Psi-Jack | Hmm | 15:25 |
| Psi-Jack | That's not even in there. | 15:25 |
| _ruben | smoser: its a intrepid upgraded to jaunty (pxe env doesnt do jaunty yet) .. prob existed in intrepid as well i think (hoped the uprade to jaunty would fix it as a "bonus") | 15:25 |
| Psi-Jack | This is from 8.04->9.04, and 9.04->9.04 it's not working for. | 15:26 |
| sgsax | Psi-Jack: see man sshd_config to find the default | 15:26 |
| Psi-Jack | Yeah, still no fixing it, so far. | 15:27 |
| sgsax | default path should be the same on both releases | 15:27 |
| Psi-Jack | Yep. | 15:27 |
| sgsax | check in /var/log/auth, see if anything shows up there | 15:27 |
| sgsax | or wherever your auth logs show up | 15:27 |
| Psi-Jack | Oh. Suddenly NOW it starts working. | 15:28 |
| sgsax | heh | 15:28 |
| Psi-Jack | Oh one system anyway. | 15:29 |
| === PhotoJim is now known as PhotoJim_ | ||
| Psi-Jack | Two systems stil not working with. heh, which is stupidly odd. | 15:32 |
| sgsax | I suppose this goes without saying, but if you modified sshd_config on any host, then you need to restart the sshd service | 15:33 |
| Psi-Jack | Well, yes, of course. | 15:34 |
| sgsax | look in their logs, see if anything helpful turns up | 15:34 |
| sgsax | just trying to cover all bases | 15:34 |
| Steve[mbp] | Morning everyon! | 15:35 |
| Psi-Jack | Oh.. | 15:35 |
| Psi-Jack | I wonder.. | 15:35 |
| Psi-Jack | it's the eCryptFS! | 15:35 |
| sgsax | hiya Steve[mbp], you must get your coffee early, either that or just naturally cheerful in the morning :) | 15:36 |
| Psi-Jack | Heh | 15:36 |
| Psi-Jack | Yeah, eCryptFS on the home directory. | 15:36 |
| Psi-Jack | Now, the question is, How the frack do I get just the authorized_keys into it non-encrypted? | 15:36 |
| sgsax | so the hosts that aren't doing key auth don't have cryptfs loaded? | 15:36 |
| Psi-Jack | yes | 15:36 |
| Psi-Jack | The reason one started working suddenly, is because I was ssh'ing in a second time. | 15:37 |
| Psi-Jack | So the cryptfs home was mounted. | 15:37 |
| sgsax | ah | 15:37 |
| Psi-Jack | Heh | 15:37 |
| Psi-Jack | So, I found one bug in the cryptfs home method! | 15:38 |
| sgsax | well done! | 15:38 |
| ball | What's a Nagios? | 15:38 |
| Psi-Jack | google it. | 15:38 |
| Psi-Jack | You'll learn more. | 15:38 |
| sgsax | ball: centralized host monitoring system | 15:38 |
| ball | Ah okay. | 15:39 |
| sgsax | http://nagios.org/ | 15:39 |
| sgsax | very flexible, very stable | 15:39 |
| sgsax | it's what most shops use | 15:39 |
| acalvo | sgsax: we've been using horde in my university for a long time | 15:40 |
| acalvo | _ruben: I really like roundcube | 15:40 |
| acalvo | but I don't think there's a package for ubuntu | 15:40 |
| acalvo | (at least not the last time I've searched for it) | 15:40 |
| acalvo | and sgsax, Zimbra is a collaborative suite, right? | 15:40 |
| sgsax | yep | 15:40 |
| sgsax | currently owned by yahoo | 15:41 |
| Psi-Jack | Ugh. | 15:42 |
| sgsax | new IT management wanted something Exchangey but without Exchange | 15:43 |
| acalvo | does horde offers the same? | 15:43 |
| sgsax | don't think so, but roundcube does, to some extent | 15:43 |
| acalvo | mmm nice then | 15:43 |
| sgsax | horde is just webmail frontend to imap | 15:44 |
| jmedina | how roundcube compare to exchange? | 15:44 |
| acalvo | isn't roundcube the same? | 15:44 |
| jmedina | there is horde groupware with more collaboration tools | 15:44 |
| sgsax | hrm, guess I haven't looked at roundcube in a while, I thought it also had collab tools | 15:45 |
| jmedina | and horde webmail.. | 15:45 |
| Psi-Jack | OpenExchange is pretty much one of the best. | 15:45 |
| acalvo | in the open-source world, I realized everyone thinks X is the best | 15:46 |
| Psi-Jack | I said, one of! | 15:46 |
| Psi-Jack | Not THE | 15:46 |
| acalvo | well, it was an opinion | 15:46 |
| jmedina | you really need to try at least 3 solutions and prepare a demo with real mail users.... | 15:46 |
| Psi-Jack | Precisely. | 15:47 |
| sgsax | yeah, that happened by committee here | 15:47 |
| Psi-Jack | Bang for buck, More of my clients had been happier with OpenExchange, and were more than willing to pay for it once they saw it in action. | 15:47 |
| jmedina | normal users use mail systems different than admins and they also have different requirements | 15:47 |
| acalvo | I'm trying to configure a MTA with IMAP and a webclient, and I'm really confused with such a bunch of options there are out there | 15:47 |
| sgsax | the rest of us just got stuck with whatever the committee decided on | 15:47 |
| sgsax | my big problem is that was aren't even hosting it on campus | 15:47 |
| sgsax | we're paying yahoo to host it for us | 15:48 |
| acalvo | (I've just found Citadel, which should cover all my expectations...) | 15:48 |
| Psi-Jack | Citadel is ... Okay... Not great but okay.. | 15:48 |
| jmedina | yea... | 15:48 |
| acalvo | I'll keep with postfix+dovecot | 15:48 |
| acalvo | and all the headaches I'm having | 15:48 |
| jmedina | :) | 15:48 |
| sgsax | the biggest problem here was that most users were using horde for their primary email client | 15:48 |
| sgsax | and it just couldn't keep up with the load, got bogged down real bad | 15:49 |
| sgsax | try throwing 20K users at horde for 12 hrs a day on a single server | 15:49 |
| sgsax | people were always complaining about lag and downtime, when it was just horde buckling under the weight | 15:50 |
| acalvo | well, for 1,5k users on my old server, it work quite well | 15:50 |
| sgsax | I'm sure it does | 15:51 |
| sgsax | those of us who use a desktop mail client rarely had problems | 15:51 |
| sgsax | personally, I only use a webmail client if I have to | 15:51 |
| sgsax | because I think they all kinda suck | 15:51 |
| acalvo | well, I don't grant my users to have access to the mail thru IMAP/POP3 outside the lan | 15:52 |
| acalvo | so, they're stuck with a webmail | 15:52 |
| sgsax | ah, that would be a problem | 15:52 |
| acalvo | oh, and they complain | 15:52 |
| acalvo | but they also complain when the printer does not print, and they do it before checking it it's switched on... | 15:53 |
| acalvo | s/it/if | 15:53 |
| Psi-Jack | Hmmm. | 16:23 |
| Psi-Jack | Are ubuntu's pre-packaged nagios extensions extensive or minimal? | 16:24 |
| jmedina | Psi-Jack: you mean nagios plugins? | 16:26 |
| Psi-Jack | plugins, yes. | 16:26 |
| jmedina | in main there is nagios-plugins-base with base plugins | 16:28 |
| jmedina | check the file list http://packages.ubuntu.com/jaunty/amd64/nagios-plugins-basic/filelist | 16:28 |
| Psi-Jack | Hmm, basic, yeah.. | 16:29 |
| Psi-Jack | Most of the very basics. Nothing for postgresql or mysql checks though. | 16:30 |
| jmedina | :) | 16:30 |
| jmedina | you can help packaging those plugins | 16:30 |
| Psi-Jack | Aight, guess I'll add those manually then. | 16:30 |
| Psi-Jack | Heh yea.. When I re-learn dpkg all over again. | 16:30 |
| jmedina | and help other future users | 16:30 |
| Psi-Jack | I haven't used deb/ubunutu in years.. | 16:31 |
| jtimberman | mathiaz: updated packages yesterday/last night with some more goods. | 16:31 |
| jmedina | there is nagios-plugins-extra which only has check_fping and check_game... | 16:31 |
| Psi-Jack | Heh yeah. | 16:32 |
| Psi-Jack | WTF is check_game? LOL | 16:32 |
| Psi-Jack | This plugin tests game server connections with the specified host. | 16:32 |
| Psi-Jack | Oh.. My... I suddenly feel stupider by looking into that. | 16:32 |
| mathiaz | jtimberman: yop - reviewing them now | 16:33 |
| jtimberman | thanks! | 16:34 |
| Psi-Jack | Well, hmm, it has check_pgsql, but it's very basic looking so far. | 16:46 |
| Sam-I-Am | mathiaz: you around? | 18:34 |
| mathiaz | Sam-I-Am: yop | 18:34 |
| Sam-I-Am | mathiaz: can you tell me what inside the openldap deb src package... or any package for that matter... determines which files from the original source tree get dumped into debian/build ? | 18:35 |
| Sam-I-Am | i'm trying to get nssov moved under debian/build... but can't seem to figure out whats handling that | 18:36 |
| Sam-I-Am | i'm also troubleshooting that nssov load failure... dynamic lib loader claims file not found, but its there... thinking the lib is borked. | 18:36 |
| Sam-I-Am | i noticed in nssov.la that the libpath points to /usr/local/libexec/openldap instead of /usr/lib/ldap ... not sure how much it matters, but its something i'm going to change before rebuilding | 18:37 |
| mathiaz | Sam-I-Am: the builddir is set in debian/rules | 18:38 |
| mathiaz | Sam-I-Am: right - I'd first focus on getting the nssov building fixed | 18:38 |
| === Faust-C2 is now known as Faust-C | ||
| Sam-I-Am | yeah, which is debian/build for openldap... but how does it know which files from the source tree to move in there? in other words, i see it copying server, client, doc, etc... but not contrib... which probably explains why nssov is getting built outside of the tree | 18:38 |
| mathiaz | Sam-I-Am: moving the nssov build to debian/builddir can be looked at the following step | 18:38 |
| Sam-I-Am | sure | 18:39 |
| Sam-I-Am | i'm just curious how stuff gets in there... | 18:39 |
| Sam-I-Am | i'm planning to fix the loading issue first | 18:39 |
| mathiaz | Sam-I-Am: right - contrib are not integrated in the main build process | 18:39 |
| Sam-I-Am | i dont see a bug for it btw... wondernig if i should create one and assign myself to it | 18:39 |
| mathiaz | Sam-I-Am: sure | 18:39 |
| mathiaz | Sam-I-Am: look a the configure-stamp target in debian/rules | 18:40 |
| Sam-I-Am | mmkay | 18:40 |
| mathiaz | Sam-I-Am: It seems that the upstream build scripts/infrastructure already supports building out-of-the tree | 18:41 |
| Sam-I-Am | ok, then we'd have to manually clean the temp build files because dh_clean won't hit it, right? | 18:42 |
| mathiaz | Sam-I-Am: yes | 18:42 |
| mathiaz | Sam-I-Am: that seems the best option for now | 18:42 |
| Sam-I-Am | ok, i wrote a patch for that already | 18:42 |
| Sam-I-Am | for debian/rules | 18:42 |
| mathiaz | Sam-I-Am: and the build infrastructure from openldap doesn't support the contrib/ | 18:42 |
| Sam-I-Am | k | 18:43 |
| mathiaz | Sam-I-Am: this is probably an Ubuntu specific patch | 18:43 |
| Sam-I-Am | yes, since deb doesnt do nssov | 18:43 |
| mathiaz | Sam-I-Am: as Debian doesn't build the nssoverlay | 18:43 |
| mathiaz | Sam-I-Am: to submit it to Ubuntu I'd suggest to use the bzr package branch | 18:43 |
| Sam-I-Am | ok, i havent done that before | 18:45 |
| mathiaz | Sam-I-Am: https://wiki.ubuntu.com/DistributedDevelopment/Documentation/ | 18:46 |
| mathiaz | Sam-I-Am: ^^ this is the documentation | 18:46 |
| mathiaz | Sam-I-Am: it's still a work in progress but should cover the whole workflow | 18:46 |
| mathiaz | Sam-I-Am: if you run into problems, ask me or james_w (in #ubuntu-devel) | 18:46 |
| Sam-I-Am | sure, no prob | 18:46 |
| mathiaz | Sam-I-Am: that will help us flesh the documentation | 18:47 |
| mathiaz | Sam-I-Am: that will help us flesh out the documentation | 18:47 |
| Sam-I-Am | i've been troubleshooting a prod issue all day so it'll be a bit before i can get back to the nssov thing | 18:47 |
| Sam-I-Am | kinda sucks when your VMs are getting maybe 6 mbit/s disk io | 18:47 |
| luckyone | hello all - can anyone help me figure out why I have a really high io wait (wa%) listed in top? | 19:29 |
| luckyone | I want to find the offending process, if possible | 19:30 |
| giovani | luckyone: sure, I can help with that | 19:30 |
| luckyone | giovani: thanks! is there a way to sort top by wa%? | 19:30 |
| giovani | wait time isn't measured per-process | 19:30 |
| giovani | only total | 19:30 |
| luckyone | some of the threads I have read point to incorrect RAID setup, but none were specific on what needs to be set | 19:31 |
| giovani | can you run a "ps aux" and pastebin the output? | 19:31 |
| luckyone | sure | 19:31 |
| giovani | sudo ps aux that is | 19:31 |
| giovani | (or run in a root shell, either way) | 19:31 |
| luckyone | sure thing | 19:32 |
| luckyone | opening a new, non-screen ssh connection... | 19:32 |
| luckyone | (this takes forever now...) | 19:32 |
| luckyone | so long in fact, ssh times out | 19:32 |
| giovani | yeah, to be expected | 19:32 |
| giovani | do you have an existing session? | 19:32 |
| luckyone | I do, I can't virtually scroll up in my screen session | 19:33 |
| luckyone | command is running | 19:33 |
| giovani | you can | 19:33 |
| luckyone | will pastebin in just a minute | 19:33 |
| giovani | ctrl-a [ | 19:33 |
| giovani | will enter scrollback mode | 19:33 |
| luckyone | right, you can - I can't (forgot how) | 19:33 |
| giovani | ^^ | 19:34 |
| uvirtbot | giovani: Error: "^" is not a valid command. | 19:34 |
| giovani | why not just redirect the output to a file, and then less, or pipe directly to less | 19:34 |
| giovani | either way will work | 19:34 |
| luckyone | yeah, totally | 19:35 |
| giovani | the bottom line with really high io wait time | 19:35 |
| giovani | is that either the disk is nearly dead | 19:35 |
| giovani | or HEAVILY fragmented | 19:35 |
| luckyone | http://pastie.org/588833 | 19:36 |
| giovani | or, you're simply doing really heavy i/o operations, and very low cpu operations -- so the ratio gets out of balance, and the cpu ends up waiting for the io operations | 19:36 |
| giovani | (or you have some fringe case where it's a bad driver, etc) | 19:36 |
| luckyone | well, the disk the OS is using is a CF drive running an ext2 filesystem | 19:36 |
| giovani | heh | 19:37 |
| luckyone | but, this is a recent problem - post upgrade to jaunty | 19:37 |
| giovani | sounds like a failing cf | 19:37 |
| luckyone | crap, I hope not | 19:37 |
| giovani | cfs have really short wite lifespans | 19:37 |
| luckyone | like how short? | 19:37 |
| giovani | like, each sector can be written max a few thousand times | 19:37 |
| giovani | which is why you generally don't put actively changing filesystems on them | 19:38 |
| giovani | i.e. filesystems where logs are written | 19:38 |
| luckyone | right - that was why I put ext2 on it | 19:38 |
| giovani | right ... but if you're writing a log file once per second ... | 19:38 |
| giovani | you'll quickly burn up those sectors | 19:38 |
| luckyone | yeah... | 19:38 |
| luckyone | damn.. | 19:38 |
| giovani | I can't be sure that it's the cause | 19:39 |
| giovani | but it's the most likely | 19:39 |
| luckyone | ok | 19:39 |
| giovani | I can run some tests with you | 19:39 |
| luckyone | sure thing | 19:39 |
| giovani | first run "vmstat 5 5" | 19:39 |
| luckyone | do they make SSD drives that are compact flash form factor? | 19:39 |
| giovani | then run "iostat" | 19:39 |
| Vog | Yep that happened to me when I used a cf to store logs on a firewall box... | 19:39 |
| giovani | luckyone: no ... that's too tiny | 19:39 |
| luckyone | dang it | 19:39 |
| giovani | luckyone: they do have microdrives | 19:39 |
| giovani | which are super-tiny hard drives in the CF formfactor | 19:40 |
| luckyone | low power too? | 19:40 |
| giovani | relatively, yes | 19:40 |
| giovani | what is the box, if you don't mind me asking | 19:40 |
| luckyone | MSI Wind | 19:40 |
| giovani | I mean, what's its function? | 19:40 |
| luckyone | it is a 1 TB RAID 1 nas | 19:40 |
| giovani | and why are you using a CF? | 19:40 |
| luckyone | so the drives can completely spin down when not active | 19:40 |
| giovani | why is the CF formfactor so important? | 19:41 |
| giovani | what would be bad about using some tiny, cheap IDE/SATA drive for the OS? | 19:41 |
| luckyone | I am out of room for regular drives | 19:41 |
| luckyone | I have a CF slot on mobo | 19:42 |
| giovani | ohh, wait | 19:42 |
| giovani | the wind | 19:42 |
| giovani | this is a netbook | 19:42 |
| luckyone | nettop | 19:42 |
| giovani | que? | 19:42 |
| luckyone | giovani: http://www.newegg.com/Product/Product.aspx?Item=N82E16856167037 | 19:44 |
| giovani | ah ok | 19:44 |
| luckyone | the older version... I have the N270 atom chipset | 19:44 |
| luckyone | not the 330 | 19:44 |
| giovani | ok | 19:44 |
| giovani | did you run the commands I asked? | 19:45 |
| luckyone | http://pastie.org/588863 | 19:47 |
| Clas | Trying to install Ubuntu 9.04 server 64bit version but installer stops and unable to install grub to hd0, centos installed grub ok but i dont like centos, anyone with some hints? | 19:48 |
| Clas | if i in the installer change to lilo, lilo installs ok | 19:49 |
| cemc | hi. is it possible to boot from an ubuntu cd but to do a netinstall? I mean to get the packages from the network instead of the cd ? | 19:50 |
| Clas | Its a clean install with default partitions | 19:50 |
| luckyone | giovani: http://pastie.org/588863 | 19:50 |
| giovani | luckyone: yeah ... your iowait isn't incredibly high, honestly | 19:51 |
| giovani | I mean ... 10-20% shouldn't make the box unusable | 19:51 |
| giovani | unless the disk is dying | 19:51 |
| giovani | which it probably is | 19:51 |
| luckyone | son of a | 19:51 |
| luckyone | the cf drive is only like .75 years old | 19:52 |
| giovani | cemc: cemc yep -- http://archive.ubuntu.com/ubuntu/dists/jaunty/main/installer-i386/current/images/netboot/mini.iso | 19:52 |
| giovani | luckyone: if you have system logs writing to it ... that's not old | 19:52 |
| giovani | CF drives are very very cheap | 19:52 |
| giovani | they're designed for write-once read-many storage | 19:53 |
| giovani | Clas: this is a grub error, unfortunately -- it's sometimes caused by a misunderstanding of which drive hd0 is | 19:54 |
| giovani | Clas: which install cd did you use? | 19:54 |
| Clas | giovani ubuntu-9.04-server-amd64.iso | 19:56 |
| Clas | in patitioning the drive i just chose guded with lvm | 19:56 |
| Clas | and accepted the default | 19:57 |
| giovani | Clas: do you have multiple hard drives in the machine you're installing on? | 19:57 |
| Clas | giovani its a promise sata raid with one 1.5 tb raid5 array made from 4 500gb disks | 19:59 |
| luckyone | giovani: if you had a cf port available, what would you put in it as your disk for / | 20:01 |
| RoyK | some 8GB CF card would do well | 20:02 |
| giovani | RoyK: nope ... we've already discussed this | 20:02 |
| RoyK | depending on how much stuff you need installed, obviously, and how much swap | 20:02 |
| RoyK | ah | 20:02 |
| giovani | luckyone: I'd use a microdrive if the CF slot was my only option | 20:02 |
| RoyK | giovani: sorry :) | 20:02 |
| giovani | RoyK: you clearly haven't worked with CFs much -- they're not useful for constantly written filesystems | 20:02 |
| giovani | like a typical *nix / | 20:02 |
| RoyK | giovani: microdrives aren't in production anymore | 20:03 |
| Clas | as it works with lilo is it ok to use it instead of grub or will i have troubble later on? | 20:03 |
| RoyK | also, CFs are quite usable for a rootfs | 20:03 |
| RoyK | there is built-in wear-leveling on all the recent (3+ years old) ones | 20:03 |
| RoyK | also, if in panic, mirror two | 20:03 |
| giovani | RoyK: he has one slot, period | 20:04 |
| giovani | no other disk options | 20:04 |
| RoyK | then use one - the chance for an error in other stuff than logs is minimal | 20:04 |
| RoyK | only the logs and perhaps tmp are written to | 20:04 |
| giovani | and I'm not aware of any decent wear-leveling in use on standard consumer CF cards | 20:04 |
| RoyK | what is this? a file server? | 20:04 |
| giovani | RoyK: he has what I assess to be a failing CF disk right now | 20:04 |
| giovani | this is a file server, with the root partition stored on the CF | 20:05 |
| cemc | giovani: thanks | 20:05 |
| RoyK | with heavy logging, and with the logs on the root, lots of stuff will be written | 20:05 |
| RoyK | I'd do network syslog and no local for such a task | 20:06 |
| giovani | exactly | 20:06 |
| RoyK | or at least nothing like debug/info/notice | 20:06 |
| giovani | or use ramdisk for it | 20:06 |
| RoyK | yeah | 20:06 |
| RoyK | erm | 20:06 |
| luckyone | ramdisk - link plz | 20:06 |
| RoyK | that is - ramdisk + udp | 20:06 |
| RoyK | ramdisk alone seems a little waste | 20:06 |
| RoyK | luckyone: mke2fs /dev/ram0 && mount /dev/ram0 /var/log | 20:08 |
| RoyK | for instance | 20:08 |
| giovani | might want to look at tiny core linux | 20:08 |
| RoyK | you can set the max size, normally 64 megs, as a kernel option or an option upon loading the kernel module | 20:08 |
| RoyK | although ramdisk is hardlinked in ubuntu, so you need a boot option | 20:08 |
| RoyK | see ramdisk= in man bootparam | 20:09 |
| AdamDV | Hello, anyone active? | 20:10 |
| AdamDV | I am wondering how to backup user mail using maildirs with postfix and dovecot? | 20:10 |
| luckyone | the machine has tons of ram for what it is being used for | 20:11 |
| luckyone | this is not recommended - http://www.memorysuppliers.com/hi4gbmi4cofl.html | 20:13 |
| henkjan | 20:14 | |
| AdamDV | 20:14 | |
| luckyone | 20:14 | |
| RoyK | 20:14 | |
| AdamDV | 20:14 | |
| RoyK | nice price - $45 for 4GB | 20:15 |
| luckyone | better option? | 20:15 |
| RoyK | a good sandisk card | 20:16 |
| RoyK | and reduced amount of writes to the root | 20:16 |
| RoyK | put /tmp and /var/log on ramfs | 20:16 |
| RoyK | for a fileserver, those two are the only ones written to | 20:16 |
| RoyK | except /home and /root when someone logs in | 20:17 |
| RoyK | and they can be other places | 20:17 |
| RoyK | put /home on something large, keep /root as it is, on the root, but don't use it | 20:17 |
| RoyK | if found of putting source code under /usr/src, put that on spinning drives | 20:18 |
| RoyK | setup network syslog (*.* @somehost in /etc/syslogd.conf) and allow just critical stuff to be written locally, perhaps also warnings, but then to a ram drive | 20:18 |
| giovani | AdamDV: it's no different than backing up other files | 20:19 |
| giovani | maildir uses directories and files ... back them up however you'd like | 20:19 |
| RoyK | luckyone: wouldn't this be quite sufficient? | 20:19 |
| AdamDV | giovani: So, I can just copy off the releavant /home/user/mail folders | 20:19 |
| AdamDV | and then copy them back? | 20:19 |
| luckyone | ok, so - I should buy a new flash drive, then move my partitions so that only extremely static things are on / | 20:19 |
| giovani | AdamDV: you mean in the event of an outage? | 20:19 |
| giovani | why would you copy files back unless the originals were gone? | 20:19 |
| AdamDV | I'm re-installing, and need to backup mail. | 20:20 |
| giovani | oh, yes | 20:20 |
| RoyK | luckyone: that's my choice - get something from sandisk - that's good stuff | 20:20 |
| giovani | just copy them | 20:20 |
| AdamDV | That would work then? | 20:20 |
| RoyK | a little more expensive, but good, and fast | 20:20 |
| giovani | AdamDV: as long as you copy everything, yes | 20:20 |
| luckyone | RoyK: will you shoot me a link or an email of what /etc/fstab should look like? | 20:20 |
| AdamDV | I see. | 20:20 |
| AdamDV | Okay, thanks. | 20:20 |
| RoyK | luckyone: make sure to mkfs those ram drives you need at boot time (/etc/init.d/something). after that, just mount /dev/ram0 /some/place (or ram1 or ram2....) | 20:21 |
| RoyK | set ramdisk size to the size you want (man bootparam) | 20:22 |
| RoyK | the ramdisk size isn't allocated statically like it was in the old days, it's just copy-on-write as always | 20:22 |
| luckyone | I may just by another box to be my NAS.... | 20:22 |
| RoyK | luckyone: try opensolaris :) | 20:22 |
| RoyK | that stuff rocks | 20:23 |
| giovani | uh | 20:23 |
| RoyK | at least for storage | 20:23 |
| giovani | please don't use zfs | 20:23 |
| giovani | it's not stable | 20:23 |
| RoyK | it's rock fucking stable | 20:23 |
| giovani | it's absolutely not | 20:23 |
| RoyK | I have a dozen servers running 24x7 on ZFS | 20:23 |
| giovani | I have a friend running a 100TB+ farm | 20:23 |
| RoyK | it's _really_ stable | 20:23 |
| RoyK | well, we only have some 70TB or so | 20:23 |
| RoyK | not a hitch | 20:23 |
| RoyK | giovani: what happened to him? | 20:24 |
| RoyK | were using RAIDZ and one drive failed and then checksum on another while rebuilding? :) | 20:24 |
| giovani | he used raidz as well | 20:24 |
| RoyK | /that's/ quite common, but not really ZFSs fault | 20:24 |
| giovani | he had some corruption issues | 20:25 |
| giovani | he runs a backup company | 20:25 |
| giovani | that wasn't acceptable | 20:25 |
| giovani | he moved back to Areca RAID cards | 20:25 |
| giovani | which are rock-solid | 20:25 |
| RoyK | usually that's what happens - one drive goes down, replace the drive, rebuild, there's an error | 20:25 |
| giovani | yep, and that might be fine for generated data | 20:25 |
| RoyK | Areca doesn't tell you about data loss with those sort of failures | 20:25 |
| giovani | but not for critical backup data | 20:25 |
| RoyK | normal RAID systems just corrupt the file | 20:26 |
| giovani | RoyK: if you're checksuming your files you know | 20:26 |
| RoyK | raidz/zfs tells you | 20:26 |
| giovani | he's found them to be rock-solid | 20:26 |
| giovani | I trust his opinion -- since he has a very large setup | 20:26 |
| RoyK | that's why you should use raidz2 | 20:26 |
| giovani | 2 racks full | 20:26 |
| RoyK | similar to raid6 with checksums | 20:26 |
| RoyK | two racks for 100TB? | 20:26 |
| RoyK | I get 20TB into 4U with RAIDZ2 | 20:27 |
| giovani | I said 100+ | 20:27 |
| RoyK | including a root and some SSDs for caching | 20:27 |
| luckyone | do you guys recommend a cheap NAS | 20:27 |
| giovani | I'm not sure of his total size right now | 20:27 |
| luckyone | standalone device? | 20:27 |
| giovani | it's probably nearing 200TB | 20:27 |
| RoyK | luckyone: I recommend lots of drives, a fast SSD or two and opensolaris | 20:27 |
| giovani | luckyone: what do you mean? a commercial product? or a platform for you to put linux on or what? | 20:27 |
| RoyK | giovani: in my setup, 200TB fits into 40U, one rack :) | 20:27 |
| RoyK | erm | 20:28 |
| RoyK | yes | 20:28 |
| giovani | RoyK: using what size drives? this wasn't built yesterday, it's grown over 5 years | 20:28 |
| luckyone | commercial product, doesn't need to run linux | 20:28 |
| giovani | luckyone: depends on your budget | 20:28 |
| luckyone | < $200 | 20:28 |
| luckyone | RAID 1 | 20:28 |
| luckyone | I have the drives | 20:28 |
| giovani | no idea | 20:28 |
| RoyK | giovani: opensolaris with raidz2 on supermicro boxes, 20TB in 4U, usable size | 20:29 |
| giovani | RoyK: I asked what size disks ... | 20:29 |
| RoyK | 1,5TB | 20:29 |
| giovani | ok, so there you go | 20:29 |
| giovani | that's a MODERN drive | 20:29 |
| giovani | his business has grown over 5 years | 20:29 |
| giovani | he doesn't just drop all old storage | 20:29 |
| giovani | so the older stuff is less dense, obviously | 20:29 |
| RoyK | since there are 15 _data_ drives in a RAID set, this thing will be FAST even if the drives are just 7200rpm | 20:29 |
| RoyK | also the SSD caching in ZFS is quite fancy | 20:30 |
| giovani | RoyK: who's arguing speed? of course it's fast | 20:30 |
| giovani | you're arguing about what fits in a rack | 20:30 |
| giovani | which is a product of the size of the drives more than anything else | 20:30 |
| giovani | which has changed drastically over the years | 20:30 |
| RoyK | I just dismissed an old NetApp at work | 20:30 |
| RoyK | 4+3+3+3U with a total of 2TB usable space | 20:30 |
| giovani | good for you? heh | 20:30 |
| RoyK | :) | 20:31 |
| giovani | NetApps are overpriced, but rock-solid | 20:31 |
| error404notfound | when i upgrade a server, how do i know it needs a reboot? | 20:31 |
| giovani | and if storage isn't your core-competency ... | 20:31 |
| RoyK | giovani: ZFS is rock solid as well | 20:31 |
| giovani | error404notfound: if you installed a new kernel | 20:31 |
| RoyK | it really is | 20:31 |
| giovani | RoyK: that's not everyone's experience | 20:31 |
| RoyK | but using raidz is not good | 20:31 |
| giovani | anyway, if storage isn't your core competency | 20:31 |
| RoyK | raidz2 can take a two-drive-failure | 20:31 |
| giovani | it's best to outsource to a company that does it for a living | 20:31 |
| error404notfound | giovani, hmmm, and what if i dont reboot? | 20:31 |
| RoyK | using raidz is NOT good | 20:31 |
| giovani | error404notfound: then you don't run the new kernel you installed | 20:32 |
| giovani | for example, my company | 20:32 |
| error404notfound | giovani, thats not a problem, uptime is what i need :P | 20:32 |
| giovani | is not full of storage experts | 20:32 |
| error404notfound | to brag :D | 20:32 |
| RoyK | giovani: I bet you what happened was your friend used raidz (single parity), one drive failed, and during rebuild, things fucked up | 20:32 |
| giovani | error404notfound: that's a shame -- you're probably missing out on security fixes | 20:32 |
| error404notfound | giovani, :D | 20:32 |
| error404notfound | kidding | 20:32 |
| giovani | RoyK: I don't know the details, I'm betting not, but there's no point in betting | 20:32 |
| RoyK | giovani: no single-parity-solution can do that | 20:32 |
| giovani | of course not | 20:33 |
| giovani | let's stop -- you're trying to guess | 20:33 |
| RoyK | well, Sun is using raidz in their NetApp-priced solutions | 20:33 |
| RoyK | Sun is using opensolaris in that | 20:33 |
| giovani | good for them, they're not NetApp | 20:33 |
| RoyK | or something similar | 20:33 |
| RoyK | giovani: give me the details of how things broke and why it is Sun's fault and I might beleive you - just saying "it's not stable, but I don't know the details" is outright stupid | 20:34 |
| giovani | it's not stupid -- it's offering the fact that alternate experiences exist, without having the details | 20:34 |
| giovani | if you don't want to listen, by all means ... don't | 20:34 |
| RoyK | please | 20:36 |
| RoyK | if I tell you ubuntu is worthless because a friend of mine had a server crash and I can't come up with any details, wouldn't that be a rather stupid thing for me to say? | 20:37 |
| Faust-C | i wonder why no one can learn to accept others differences w/o resorting to name calling etc | 20:37 |
| Faust-C | RoyK, actually everyone is quite aware of opensolaris' instability, even Sun themselves | 20:37 |
| RoyK | Faust-C: strange thing they use it for their storage solutions, then | 20:38 |
| Faust-C | their not banking on stability as much as they are banking on just trying to get people to switch because its new and innovative and does have some benefits compared to other solutions | 20:38 |
| Faust-C | RoyK, show me any solution that is fail proof | 20:38 |
| Faust-C | there is none | 20:38 |
| RoyK | Faust-C: can you document opensolaris' instability? | 20:38 |
| Faust-C | windows is highly unstable yet ppl use it everyday | 20:38 |
| Faust-C | RoyK, no but i can provide you with resources in regards to that | 20:38 |
| Faust-C | their wiki, forum and irc room all are based around ISSUES | 20:39 |
| RoyK | Faust-C: for instance, Solaris 10 has nothing like dedup and in-kernel cifs. in-kernel cifs is in opensolaris, and dedup comes in february | 20:39 |
| RoyK | Faust-C: and ubuntu and ext-fucking-3 is so much better than zfs? gimmi a break | 20:39 |
| RoyK | I know linux pretty good, having worked with it daily for 10 years | 20:39 |
| Faust-C | does it really matter which is better or matter what means more to you | 20:40 |
| Faust-C | if it meets your needs then ok, if not oh well | 20:40 |
| RoyK | don't misunderstand - we have more linux servers than solaris servers | 20:40 |
| luckyone | I can get a microdrive for 20.00 | 20:40 |
| RoyK | I'm just saying that for storage, opensolaris is way ahead of linux | 20:40 |
| Faust-C | RoyK, i have found areas that linux servers havent been able to fill | 20:40 |
| Faust-C | RoyK, and i agree tbh | 20:40 |
| luckyone | would putting / on that be a simple and effective solution? | 20:40 |
| Faust-C | zfs is a amazing technology | 20:40 |
| Faust-C | and thanksfully Sun made it opensource | 20:41 |
| Faust-C | luckyone, yeah | 20:41 |
| Faust-C | luckyone, can i suggest a alternative | 20:41 |
| RoyK | luckyone: erm, I'd use an SD drive and move the writing to something like a ram drive or the data drives | 20:41 |
| Faust-C | luckyone, http://www.transcendusa.com/Products/CatList.asp?Func1No=1&FldNo=18 | 20:41 |
| RoyK | luckyone: you can use the data drive set for /var and /tmp and stuff if you don't want ram drives | 20:41 |
| Faust-C | luckyone, i used the ide flash module and i love it | 20:42 |
| Faust-C | makes OS faster while data is safe on a raid setup | 20:42 |
| RoyK | luckyone: the microdrive will also die, but when it does, it'll most likely DIE, | 20:42 |
| luckyone | gotcha | 20:42 |
| RoyK | Faust-C: he only has a CF slot available | 20:42 |
| luckyone | microdrive is not perm | 20:42 |
| Psi-Jack | Hmmm. | 20:42 |
| Faust-C | oh well they have CF cards as well | 20:42 |
| luckyone | I need to look into ramdisk | 20:43 |
| RoyK | CF cards work well | 20:43 |
| Faust-C | luckyone, keep it simple | 20:43 |
| RoyK | luckyone: and you have data drives - use them for /var and /tmp and so on | 20:43 |
| Faust-C | dont go off on a tangent, keep it simple period | 20:43 |
| giovani | luckyone: no drives are forever | 20:43 |
| Psi-Jack | I've always wondered about this, but in some many cases, lm-sensors stuff, reports that.. For example, the min fan1 speed should be -1 RPM, div = 8.. Why the heck is it -1 and what's div for? | 20:43 |
| giovani | luckyone: microdrives will fail before most normal hard drives | 20:43 |
| RoyK | or if you don't want to repartition, dd if=/dev/zero of=/storage/var bs=1M count=10k | 20:44 |
| giovani | but long after a CF if you're writing to it | 20:44 |
| RoyK | and mkfs on that one | 20:44 |
| RoyK | mont it on /var | 20:44 |
| RoyK | done | 20:44 |
| RoyK | giovani: most CFs today can take 100k write ops | 20:44 |
| RoyK | _per_sector_ | 20:44 |
| RoyK | with wear leveling, that means _for_years_ | 20:44 |
| RoyK | unless you use it for something doing very heavy logging | 20:45 |
| Faust-C | wow | 20:45 |
| RoyK | but don't use heavy logging | 20:45 |
| Faust-C | i didnt know that, thanks for the info RoyK | 20:45 |
| Faust-C | use network logging server | 20:45 |
| RoyK | yeah, and perhaps log WARN and CRIT and so on to a file | 20:45 |
| Psi-Jack | rsyslog FTW! | 20:45 |
| RoyK | *.* @logserver in syslog.conf | 20:46 |
| RoyK | syslogd -r | 20:46 |
| Psi-Jack | @@ would be better. | 20:46 |
| RoyK | :þ | 20:46 |
| Psi-Jack | UDP is so unreliable. :p | 20:46 |
| RoyK | why? | 20:46 |
| RoyK | not really on LANs | 20:46 |
| RoyK | is @@ over tcp? | 20:46 |
| Psi-Jack | Yes, really, even on LANs. | 20:46 |
| Psi-Jack | Yep | 20:46 |
| RoyK | not all syslogd versions support tcp, though | 20:47 |
| luckyone | gentlemen, thanks for the info - I have to go hop on a demo | 20:47 |
| Psi-Jack | RoyK, So, use only those that do. D'uh. Or even better use RELP. | 20:47 |
| luckyone | I will be back later to figure out how to not be so stupid | 20:47 |
| luckyone | ramdisk for logging/etc definitely makes sense | 20:47 |
| RoyK | Psi-Jack: udp works well on LANs | 20:47 |
| luckyone | because I have tons of ram | 20:47 |
| RoyK | Psi-Jack: and I don't syslog over the internet | 20:48 |
| Psi-Jack | RoyK, Well, but not perfect, and without any type of guarantee. | 20:48 |
| Psi-Jack | Yes, I have seen, even over UDP over a LAN, messages being lost. | 20:48 |
| RoyK | Psi-Jack: how much data loss would you get over a full-duplex switched network? | 20:48 |
| giovani | depends on the network obviously | 20:48 |
| giovani | and the quality of the network gear | 20:48 |
| Psi-Jack | RoyK, How heavilly loaded are the servers? How much logging occurs? | 20:49 |
| Psi-Jack | Yes, and what is the quality of the network gear, and recieving server? | 20:49 |
| RoyK | data loss on ethernet is in 99,99999999999% a full/half duplex problem or bandwidth problem, and with most networks at gigabit speed today... | 20:49 |
| RoyK | duplex problems suck | 20:49 |
| RoyK | and have always done | 20:50 |
| Psi-Jack | I even have a Gigabit network at home, but it's still possible to have issues. ;) | 20:50 |
| maswan | RoyK: go to 10GE, no half duplex there. :) | 20:53 |
| RoyK | maswan: not on 1GE either | 20:54 |
| RoyK | only 10 and 100 supports that stuff | 20:55 |
| Psi-Jack | Heh | 20:55 |
| RoyK | it came from 10 with coax, and was obsoleted with TP on 100Mbps, but kept for compatibility (for what reason???) | 20:55 |
| RoyK | I've spent some hours debugging duplex problems with 100Mbps | 20:56 |
| RoyK | it sucks rather hard | 20:56 |
| Psi-Jack | Anyway, Different topic | 20:58 |
| Psi-Jack | lm-sensors! | 20:58 |
| Psi-Jack | What's up with weirdness in some computers' sensors?! | 20:58 |
| sgsax | lack of standards and low-quality sensor chips | 20:59 |
| Psi-Jack | I've always wondered about this, but in some many cases, lm-sensors stuff, reports that.. For example, the min fan1 speed should be -1 RPM, div = 8.. Why the heck is it -1 and what's div for? | 20:59 |
| Psi-Jack | That's from an HP desktop mini-tower. | 20:59 |
| Psi-Jack | Hehe, wierdness. | 21:02 |
| Psi-Jack | Apparently this same system, people had problems using the two SATA connectors on it, because BIOS itself didn't even support SATA at all. | 21:02 |
| sgsax | nice | 21:05 |
| sgsax | I've always had a dislike for HP desktops & laptops | 21:05 |
| sgsax | compaq too, for that matter | 21:05 |
| Psi-Jack | Heh, Yeah. | 21:05 |
| Psi-Jack | I have a P3 730Mhz Compaq desktop model, I can't boot without having some kind of keyboard attached. No way to stop the error about lacking keyboard! | 21:06 |
| Psi-Jack | Tis okay though. It's just my authentication server. ;) | 21:06 |
| RoyK | Psi-Jack: usually lm-detect or whatever it's called finds several drivers that "match" and only (if lucky) one of them really work | 21:06 |
| RoyK | s/$/s/ | 21:06 |
| Psi-Jack | Heh | 21:07 |
| Psi-Jack | Great! | 21:07 |
| RoyK | sensors-detect | 21:07 |
| Psi-Jack | So sensors is about pretty welll useless in some cases, or are just plain unreliable in almost every case. | 21:07 |
| RoyK | try different versions of those that look somehow the same | 21:07 |
| RoyK | Psi-Jack: just spend some minutes to see if you can get good results from a driver. Usually you can | 21:08 |
| RoyK | but not always | 21:08 |
| Psi-Jack | Heh | 21:08 |
| Psi-Jack | I usually just ran sensors-detect, made it go through everything, hit enter, to accept, blindly, and when it's done, I run it and see what the results are. | 21:08 |
| maswan | RoyK: No, there is half duplex in GE | 21:09 |
| maswan | RoyK: but autoneg is mandatory | 21:09 |
| maswan | RoyK: very early there was actually gige hubs | 21:09 |
| RoyK | Psi-Jack: as far as I have seen, the output, including errors, has been reproducable, so you can probably trust something that works now | 21:09 |
| RoyK | wtf! | 21:09 |
| RoyK | gigabit HUBS? | 21:09 |
| Psi-Jack | Heh | 21:10 |
| RoyK | you have to be quite drunk _and_ stoned to make something like that | 21:10 |
| RoyK | from http://en.wikipedia.org/wiki/Gigabit_ethernet "Half-duplex gigabit links connected through hubs are allowed by the specification but in the marketplace full-duplex with switches are normal." | 21:10 |
| maswan | RoyK: well, gige as a standard is actually quite old, and this was from the time where a 100Mbit switch was significantly more expensive than a 100Mbit hub | 21:10 |
| RoyK | anyway - I've never seen half duplex in gigE | 21:11 |
| RoyK | I've hardly seen 100Mbps hubs as well | 21:11 |
| maswan | no, me neither, I just know that the hubs existed briefly, and it is yet another case of a thing that should have gone away | 21:11 |
| RoyK | they were quite early | 21:11 |
| RoyK | around the 100Mbps days, switches became quite normal | 21:11 |
| RoyK | last time I was using hubs in production, was back in 1996 or so while replacing a thin coax ethernet setup with a 3com-crap-hubs-but-although-twisted-pair-setup | 21:12 |
| maswan | yeah, I'm not sad to see them gone | 21:14 |
| sgsax | all lm-sensors can get of my desktop mobo is CPU core temp | 21:22 |
| Faust-C | hmm lm-senses | 21:37 |
| Faust-C | sgsax, depends on the mobo as well and if the fans support controlling etc | 21:37 |
| sgsax | personally, I'm more interested in temps than controlling fans | 21:37 |
| Psi-Jack | Yeah, fan speeds, and temps | 21:38 |
| cemc | I have a qlogic adapter on an ibm blade center, and ubuntu won't boot, it just drops me initramfs shell. any ideas? | 21:42 |
| cemc | I've downloaded the firmware, copied it in /lib/firmware and did a rmmod/modprobe qla2xxx, but it still can't find it | 21:43 |
| * jmedina needs to try ubuntu on ibm bladecenter :S | 22:02 | |
| artillerytx | Hey guys i can't figure out how to get a subdomain to work | 22:14 |
| artillerytx | sometimes it works sometimes it doesn't... any ideas? | 22:14 |
| KillMeNow | for DNS? | 22:14 |
| artillerytx | bind | 22:14 |
| KillMeNow | yea, it's easy | 22:14 |
| KillMeNow | if you are SOA for a main domain, it's just a IN A record addition | 22:15 |
| KillMeNow | so it would look like this as part of your zone record | 22:15 |
| artillerytx | well i created an A name record and then created a virtual server with that domain name | 22:15 |
| KillMeNow | subdomain IN A <IP ADDRESS> | 22:15 |
| KillMeNow | did you do a rndc reload? | 22:15 |
| KillMeNow | or /etc/init.d/bind restart ? | 22:16 |
| artillerytx | yeah in my wwmcd.org .hosts file i added swot.wwmcd.org IN A <IP ADDRESS> | 22:16 |
| artillerytx | yeah i thought i did | 22:16 |
| artillerytx | i can try the rndc reload | 22:16 |
| KillMeNow | don't need the wwmdd.org as part of your IN A record | 22:16 |
| KillMeNow | just the swot | 22:17 |
| KillMeNow | you will need to reload the bind server to populate the subdomain | 22:17 |
| artillerytx | whats the command to do that | 22:17 |
| KillMeNow | rndc reload | 22:18 |
| KillMeNow | or /etc/init.d/bind9 reload | 22:18 |
| artillerytx | it tells me rndc: connect failed 127.0.0.1#953: connection refused | 22:19 |
| artillerytx | killmenow the only place i have 127.0.0.1 in my ifconfig is for the lo | 22:21 |
| === Authority is now known as Guest3246 | ||
| uvirtbot | New bug: #413211 in lsb (main) "lsb_release crashed with ImportError in <module>() (dup-of: 383697)" [Undecided,New] https://launchpad.net/bugs/413211 | 22:23 |
| === Guest3246 is now known as Authority | ||
| uvirtbot | New bug: #413077 in ubuntu "lsb_release crashed with ImportError in <module>() (dup-of: 383697)" [Undecided,New] https://launchpad.net/bugs/413077 | 22:27 |
| uvirtbot | New bug: #415352 in lsb (main) "lsb_release crashed with ImportError in <module>() (dup-of: 383697)" [Undecided,New] https://launchpad.net/bugs/415352 | 22:27 |
| artillerytx | how do you see the owner of a file | 22:30 |
| artillerytx | ls -l | 22:31 |
| uvirtbot | New bug: #415515 in lsb (main) "após update do 9.04 para o 9.10, no boot apareceu "!" (dup-of: 383697)" [Undecided,New] https://launchpad.net/bugs/415515 | 22:31 |
| uvirtbot | New bug: #413855 in lsb (main) "lsb_release crashed with ImportError in <module>() (dup-of: 383697)" [Undecided,New] https://launchpad.net/bugs/413855 | 22:32 |
| artillerytx | So anyone know how i can fix this | 22:33 |
| KillMeNow | sorry, work called | 22:34 |
| artillerytx | Its coo | 22:34 |
| KillMeNow | ls -la will do it too Art | 22:34 |
| KillMeNow | i aliased ll to ls -la | 22:35 |
| KillMeNow | in .bashrc | 22:35 |
| artillerytx | Did you see what i said about when i run rndc | 22:35 |
| artillerytx | - rndc: connect failed: 127.0.0.1#953: connection refused | 22:35 |
| KillMeNow | so you can edit your .bashrc and add alias ll='ls -la' | 22:35 |
| KillMeNow | yea, you need to set up your config for rndc key | 22:35 |
| KillMeNow | just do the /etc/init.d/bind9 restart | 22:35 |
| artillerytx | k that worked | 22:36 |
| KillMeNow | if you google that specific error you will find TONs of how to fix that error articles | 22:36 |
| artillerytx | yeah i see a ton but im not too sure what im supposed to do i can't find that conf. file | 22:36 |
| choppy | Hello all, could use some help with a raid issue | 22:36 |
| KillMeNow | updatedb | 22:36 |
| KillMeNow | then do a locate | 22:36 |
| KillMeNow | locate <filename> | 22:36 |
| choppy | Everything with my raid works, except I don't have links in /dev/disk/by-uuid for my /dev/md0 and /dev/md0p1 devices. | 22:37 |
| KillMeNow | if it's a hardware raid I may help, don't dink much with software raids | 22:37 |
| choppy | I'm using 9.04 | 22:37 |
| choppy | any ideas? | 22:37 |
| artillerytx | i can run rndc-confgen to generate rndc.conf and copy that to named.com? | 22:37 |
| KillMeNow | i did mine by hand | 22:40 |
| KillMeNow | so ummm, yea... don't know if doing the rndc-confgen will do that | 22:40 |
| KillMeNow | sounds right tho | 22:40 |
| KillMeNow | did you google and find a walk through? | 22:40 |
| artillerytx | http://ubuntuforums.org/showthread.php?t=281393 | 22:40 |
| artillerytx | i ran rndc-confgen and its just sitting there | 22:41 |
| giovani | choppy: you don't need UUIDs for md devices | 22:43 |
| giovani | choppy: you won't be able to mount them via UUID anyhow | 22:43 |
| mathiaz | kirkland: hey - what do you think about the dh-make init script template? | 22:44 |
| mathiaz | kirkland: /usr/share/debhelper/dh_make/debian/init.d.lsb.ex | 22:44 |
| choppy | I use to mount them that way in 6.06 | 22:45 |
| mathiaz | kirkland: does it use the status code you've added to the lsb package? | 22:45 |
| giovani | choppy: there's no reason to | 22:45 |
| giovani | choppy: md devices don't work the way regular drives do where they can come up in different orders | 22:45 |
| giovani | md0 will be md0 | 22:45 |
| giovani | it's why they're being mounted that way | 22:45 |
| choppy | I was wanting to avoid a posible name overlap if a brought in another md0 from another system. | 22:46 |
| choppy | or that was my thinking, might be flawed. | 22:46 |
| choppy | I guess the kernel won't try to assemble it if it's not listed in /etc/mdadm/mdadm.conf though | 22:48 |
| choppy | Thanks for the respone giovani | 22:49 |
| Psi-Jack_ | Okay. So, if I took the deb-src packages of linux-igd from 9.04, how would I compile that into a deb for 8.04 since linux-igd was completely skipped in hardy. | 22:50 |
| Psi-Jack_ | ? | 22:50 |
| KillMeNow | ART: http://www.linuxforums.org/forum/servers/6406-rndc-connect-failed-connection-refused.html | 22:51 |
| artillerytx | I added whatever rndc-confgen gave me to /etc/bind/named.conf | 22:51 |
| KillMeNow | check that link out for the rndc connect failed | 22:51 |
| artillerytx | and rndc reload still doesn't work | 22:51 |
| Psi-Jack_ | Anyone able to help me with that? | 22:53 |
| kirkland | mathiaz: i saw it for the first time today, actually, reviewing condor | 22:58 |
| mathiaz | kirkland: and what's your opinion on it? | 23:05 |
| mathiaz | kirkland: IIUC it doesn't include your code for checking the status of a process? | 23:05 |
| kirkland | mathiaz: yeah, it crossed my mind to suggest to the maintainers to add that | 23:06 |
| uvirtbot | New bug: #416155 in bacula (universe) "package bacula-director-mysql 2.4.4-1ubuntu5 failed to install/upgrade: el subproceso post-installation script return output code error 1" [Undecided,New] https://launchpad.net/bugs/416155 | 23:11 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!