[00:00] I tend to disagree, but we'll test and find out... [00:00] jmarsden: does it add a virtualhost automaticly, or what? [00:00] (I hope not) [00:01] or at least not enable it [00:01] The default virtualhost exists "automatically". phpmyadmin adds a glibal Alias for /phpmyadmin and some config info for the directory where it puts its files, including an AddType. [00:02] I'll have a Hardy 8.04.3 server to test in about 2 more minutes... :) [00:04] hm, cherokee uses some sort of "wizards" to configure webapps, that sound like a better approach to me (it's still dead-easy to set up something like phpmyadmin, django, drupal, etc., but nothing is set up by default) [00:05] Feel free to create wishlist bugs in LP if you have a workable new approach... [00:07] sorry went to grab a sammich [00:08] per webmin, i have a default, and a virtual. [00:09] Don't use webmin, it is unsupported in Ubuntu. [00:09] !webmin [00:09] webmin is no longer supported in Debian and Ubuntu. It is not compatible with the way that Ubuntu packages handle configuration files, and is likely to cause unexpected issues with your system. See !ebox instead. [00:09] usermin? then? [00:09] !ebox [00:09] ebox is a web-based GUI interface for administering a server. It is designed to work with Ubuntu/Debian style configuration management. See https://help.ubuntu.com/community/eBox [00:12] Hmmm, I must have used the wrong ISo for my VM installm and got Ubuntu Desktop 8.04.3 instead of server... checking... [00:12] could that be the issue, that's it's desktop not server? [00:13] No, it shoudn't make a difference to how Apache works. [00:13] Although officially you shouldn't be asking for Xubuntu support in #ubuntu-server :) [00:13] didnt think so. i know i used a server install disk, but installed xubuntu cause this thing isn't very powerful. true. but i was referred here from #xubuntu >_> [00:17] questions about apache are fine here, doesn't matter much which kernel you use (except for performance under high load and certain hardware support maybe) [00:21] rootlinuxusr: OK, I just did a tasksel lamp install, added that phpingo.php file into /var/www/ and browsed to it, and it worked "out of the box". [00:21] So somehow your apache has been configured differently to mine, I would suspect... what have you done to it? [00:21] okay. so abort, retry, fail i guess. should i completely reinstall or just completely remove everything and try that? [00:22] rootlinuxusr: well, that's extreme... what did you change? [00:23] i downloaded phpmyadmin, and im assuming the error occured somewhere around here.. http://pastebin.com/d5498c8f0 [00:27] Um. Why are you manually installing some unusual version of libapache2-mod-php5 ? [00:27] because it wasn't found in the apt archives, so i downloaded it from the debian site? [00:28] Aargh. so you don't really have a tasksel lamp-server install at all [00:28] It might have been good to mention this earlier :) [00:28] actually, that's apache 1.x ? [00:29] I just installed things from the 8.04.3 CD and libapache2-mod-php5 most definitely installs just fine (by default as part of the tasksel process) [00:29] "libapache-mod-php5" I mean [00:30] JanC: Ah, yes, so it may have done nothing useful... but... in principle, this is now one very non-standard setup for no good reason. [00:30] so if libapache-mod-php5 refuses to be downloaded and installed for whatever reason, what should I do? get another disk? [00:30] rootlinuxusr: You should apt-get purge all the apache2 related things you installed, then sudo tasksel lamp-server [00:31] And it will either find it in the archives, or we will help you debug that issue. [00:31] jmarsden: I'd say all apache & apache2 related packages [00:31] What you should not do is guess and grab random stuff from Debian and mix it in and hope! [00:31] (and then not tell us you did that!) [00:32] well i thought that when i removed all the files from the website that suggests installing tasksel, that would have been removed; or for that matter Im not even entirely sure that it installed in the first place - so i ran the tasksel install method. [00:32] !lamp [00:32] LAMP is an acronym for Linux-Apache-MySQL-PHP. However, the term is often used for setups using alternative but different software, such as Perl or Python instead of PHP, and Postgres instead of MySQL. For help with setting up LAMP on Ubuntu, see https://help.ubuntu.com/community/ApacheMySQLPHP - See also the Server CD installation process (different in Edgy+) [00:33] rootlinuxusr: The last command in your pastebin is most definitely not sudo tasksel install lamp-server [00:33] So undo all the weord stuff and do that, and we'll take from there. [00:33] *wierd [00:34] well that wasnt the last from history either, but that was the last related to installing anything related to lamp [00:34] per https://help.ubuntu.com/community/ApacheMySQLPHP should i remove all of these? will that take care of things - https://help.ubuntu.com/community/ApacheMySQLPHP [00:34] err... apache2 apache2-mpm-prefork apache2-utils apache2.2-common libapache2-mod-php5 libapr1 libaprutil1 libdbd-mysql-perl libdbi-perl libmysqlclient15off libnet-daemon-perl libplrpc-perl libpq5 mysql-client-5.0 mysql-common mysql-server mysql-server-5.0 php5-common php5-mysql [00:35] rootlinuxusr: Sounds reasonable, but purge them, not just remove, so their config info goes away too. [00:36] New bug: #498907 in samba (main) "package samba-common 2:3.3.2-1ubuntu3.2 failed to install/upgrade: " [Undecided,New] https://launchpad.net/bugs/498907 [00:36] http://pastebin.com/d77b75a2a that's the result. is that deb package gone/is my install clean now? [00:38] Mostly... there is still a /etc/php5/apache2 directory apparently? Can you manually get rid of that? [00:38] Actually all of /etc/php5/ can go [00:39] gone. [00:39] OK. Now do sudo tasksel install lamp-server # and if you get any errors from it, pastebin them. [00:40] asked for pw, and had me retype - normal so far. [00:40] Good. [00:40] exited with no prompt - back to terminal's prompt [00:41] running the echo command from earlier to test. [00:41] localhost/phpinfo.php still tries to download. [00:41] How are you testing this? From the server using w3m or lynx or what? [00:42] firefox. [00:43] (I forgot you have a GUI on the "server" :) Something is odd... cached result, I am guessing... [00:43] try rebooting and see if that helps? [00:43] Stop apache2 and then run sudo htcacheclean -rv [00:44] what do the apache logs say? [00:45] when i tried htcach...(copied and pasted) it brings up the man page for it. [00:45] JanC: I have a feeling this is mod_disk_cache doing more caching than we want... [00:46] rootlinuxusr: My fault, it needs the path to the cache... let me find it... [00:46] why would that module be loaded anyway? ;) [00:47] (or used) [00:47] JanC: I don't know, but I fixed this issue for someone before and I think (memory returning as we go here) it related to this. [00:48] sudo htcacheclean -rv -p /var/cache/apache2/mod_disk_cache [00:48] same thing [00:49] http://pastebin.com/m45c286 [00:49] maybe remove the space between -p and the URL? [00:50] er, -p and teh path [00:50] same [00:51] sudo htcacheclean -rv -p /var/cache/apache2/mod_disk_cache -l1 [00:52] (It requires the -l option apparently) [00:52] okay now that did something [00:52] OK, now start apache and retest [00:52] stats size limit, total size, and total entries were all 0 [00:53] ^_^ sweet. [00:53] rootlinuxusr: Error: "_^" is not a valid command. [00:53] parses fine, thanks. iunno what the issue was apart from my errant ideas as to downloading stuff. [00:53] OK, we got it :) [00:57] jmarsden: that sounds like a bug somewhere? [00:57] JanC: Yes, and that's the second or third time I've helped someone with that now (this time I'm writing down what it was, for future reference!)... [00:58] But I don't know how to trigger the bug (over aggressive use of mod_disk_cache)... [00:59] It seems to happen if you do a "confused" apache2/php5 install the first time around, such that even fixing it later using apt-get purge won't clear the problem until you reset that cache. [00:59] why is it used in the first place? [01:00] Well, presumably it has performance benefits, why else use caching? But I'm not really sure, it just "is", in Debian/Ubuntu apache2 installations, as far as I know. [01:02] dynamic content should never be cached by default [01:03] JanC: But the bad install meant it looked like static content (download the file not run it)... that's the whole problem :) [01:04] Presumably over-caching is caused by your applicatin using incorrect HTTP fields, e.g. ETag and Pragma-Cache [01:04] twb: I doubt phpmyadmin has such obvious bugs after all those years? [01:04] twb: Well, if it is affecting just a one line .php file, that seems unlikely... [01:04] JanC: I don't trust anything that has "php" or "my" in its name :-) [01:05] @twb is there something else i should be using? [01:05] twb: I'm not a fan of "php" either, but still, I remember patching phpmyadmin in 2000, so it's been around long enough to get the obvious bugs fixed ;) [01:05] Shrug. [01:06] rootlinuxusr: I encourage administration of databases via the CLI, just like the rest of the system. [01:07] On that basis that 1) sooner or later, you're gonna need to know how to do recovery via a CLI; and 2) exposing ssh alone is better than exposing ssh AND some web-based tool. [01:07] OTOH, IANADBA. [01:08] hi, im on the process to configure a bind server..need some site that offers a step by step howto..thanks [01:08] ianadba? [01:09] I am not a DBA [01:10] twb: for 2) the alternative is to port-forward the database to your desktop through an ssh tunnel & run whatever fancy GUI tools you want ;) [01:10] JanC: true [01:10] And I'd prefer THAT over phpmyadmin, too. [01:11] ssh -fX db.debian.org oobase or whatever [01:11] Oh, sorry, like you say - mysql probably listens on ports as well as sockets by default [01:12] yeah, I just forward the mysqld port if I want to do that (no X on my server) [01:12] jongbergs: First, read the Ubuntu Server Guide, as the /topic here says... https://help.ubuntu.com/9.10/serverguide/C/dns.html has basic info on DNS setup. [01:16] I know there is another open source project besides unattended that will manage network installs of windows from Linux. Could anybody tell me the name of it? [01:21] well i totally would prefer to go CLI but im no programmer...and im just doing this to create a db with tables relating to my cards - with a hidden form to add/remove; and a form for viewing what i have so that when im at the card shop i know that i have/dont have the card...etc...or is there a OS project that already does this? [01:22] rootlinuxusr: for CCGs? [01:22] baseball preferably xD [01:22] Dunno [01:22] you don't need to be a programmer to use the CLI [01:23] true. But its a different mindset than I can get a handle around - the for dummies book stumps me... >_> [01:23] back when I started using computers all users had to use the CLI as for most of us there was nothing else (well, when I sterted to use DOS, Windows 2.x existed, but there were no applications for it, and & mac was way too expensive ;) ) [01:24] There are a bunch of generic asset-tracking utilities, but I can't easily find one geared towards card collections [01:24] JanC: Back when I started using computers they used punched cards and paper tape :) [01:24] rootlinuxusr: as you have a GUI, why not use GUI tools anyway? ;) [01:25] jmarsden: just don't drop those punched cards before numbering them ;) [01:25] well i have one team all converted to OO.O base, but I'd love to either convert that to mysql; or retype it into mysql for the speed. [01:25] OOo Base can use MySQL [01:26] sure can, but with an already existing dbase? and how could i convert that web-based so i can view without having to pull out ssh on my phone - G1 rooted. [01:28] it should be possible to use the same database for all that, but if you don't really want to keep OOo Base around, then just creating a web-based interface might be easier [01:28] i sense a great deal of googlin' ahead. [01:28] ☺ [01:28] JanC: you draw a diagonal line across the top [01:30] twb: ah, should work too indeed (a former teacher of me told me he dropped hours work of punched cards as a student once, which is what I was referrring too) [01:31] not sure if afterwards he numbered them or used a diagonal line or whatever [01:32] I heard that story 15 years ago, so don't remember all details [01:40] JanC: everyone drops cards sooner or later [01:40] A diagonal line is like journalling only your metadata -- it's not good enough, but it's fast and it's better than nothing :-) [02:13] jmarsden: if i use ns1.example.com, should i change my dns server hostname to ns1? [02:14] Noooo [02:14] ns is a role, so you make a CNAME to the machine's real name. [02:14] * twb waves around RFC1178-on-a-stick [02:15] (Also, renaming a host while it's in production can be a major pain in the arse, especially if it's running FDS and Domino.) [02:18] twb: i followed the steps in https://help.ubuntu.com/8.10/serverguide/C/dns.html, but when i run named-checkzone example.com /etc/bind/zones/example.com.db, i got this message zone example.com/IN: NS 'ns1.example.com' has no address records (A or AAAA) [02:19] jongbergs: that's saying that ns1.example.com doesn't resolve. [02:19] twb: what could be wrong? [02:19] Personally I don't use bind as a caching DNS proxy -- I prefer the simplicity of dnsmasq. [02:21] twb: i wanted to do some testing to host our school's domain name [02:22] jongbergs: oh, you want to host records for your own domain? [02:23] BTW, I notice you linked to the 8.10 document. You probably want to run the latest LTS (8.04) or the latest non-LTS (9.10). [02:34] I've set up a Ubuntu instance on Amazon Ec2 and wanted to turn it off. How do I save an image of its current state? I googled and found nothing, but something that isn't useful for me because I don't have this instance in EBS yet [02:35] How can I delete/remove a guest that's been created using virt-install? .. I've tried virsh -c qemu:///system destroy .. but all it does is shutting it down more aggressively, it doesn't remove the actual guest system. [02:41] If anyone else wonders the same thing.. use "undefine" instead of "destroy" ;) [02:55] how do i find the host name? [02:58] Not very patient that one. [03:03] Hello [03:05] is there anything i can put in ufw to block IP's that send LOTS of packets to the server? [03:10] X-M4-X, there is a script out there to do just that but the name escapes me [03:10] fail2ban, denyhosts? [03:10] it will cut off an IP that surpasses so many packets [03:10] or not, I don't think those work on just packets [03:11] I blocked the ping of death on port 7. installed denyhosts for brute force attacks upgraded my password with one to 100/100 strength [03:12] if an ip is rapidly sending packets to my server - it will ban the ip - which is the goal [03:23] Dec 20 20:26:44 emplar kernel: [ 342.020945] eth0: increased tx threshold, txcfg 0xd0f01010. <-- Does this line mean anything to you guys? [03:25] or is it pretty insignificant? [03:26] _{it looks informational. i wouldn't worry about it unless you're noticing problems with network performance
[03:26] ok, thanks
[03:27] bogeyd6: -m recent!
[03:27] ???????????/
[03:27] ///
[03:27] bogeyd6: instead of fail2ban
[03:27] kk
[03:31] aww
[03:31] http://www.debian-administration.org/articles/187
[03:31] something i did blocked connections to the server
[03:32] -j TARPIT or -j CHAOS are good, too
[03:33] But if you just want to block hosts that send lots of packets, -m limit is sufficient.
[03:35] Oops, make that hashlimit.
[03:39] looks like im gonna be reinstalling ubuntu server!
[03:39] Hooray!
[03:40] Oh thank you lord in heavan
[03:44] OMFG I FEEL RETARDED!
[03:46] newbie question regarding grub2 on karmic server RAID 1 array: if /dev/sdb fails, the system boots with the array in degraded mode. Even if grub-install is used to install grub to /dev/sdb, in case of /dev/sda failure, grub cannot bring up the array as it cant seem to find the root, so ... I found what I need to do on http://riseuplabs.org/grimoire/storage/software-raid/ ... however that is for legacy grub.
[03:47] could someone please point me to the correct steps to follow for installing grub2 to /dev/sdb so that even if /dev/sda fails, the system still boots up
[03:52] installing grub to /dev/md0 does not work
[03:52] anyone here with RAID 1 + grub 2 experience on karmic server ?
[03:56] bgoldsmith: I last tested that configuration on Lenny, where it completely failed
[03:56] #grub told me it was All Much Better NowTM
[03:57] I've had more success with extlinux, though.
[03:58] ouch ... I have to deploy a RAID server in an enterprise environment in the next week, and I've recommended that we go with Ubuntu server
[03:58] bgoldsmith: why did you recommend that?
[03:59] because - people here are familiar with ubuntu
[03:59] desktop
[03:59] and I won't be here as an admin for long (I
[04:00] have the post of 'principal engg')
[04:00] bgoldsmith: One way around this would be to install Jaunty and upgrade to Karmic (if you don't want Hardy), then you'll have legacy grub.
[04:00] hmmm
[04:00] I was thinking of doing that
[04:01] but for lack of time, I was wondering if there was a GRUB2+RAID1 howto somewhere google couldn't reach
[04:01] Or just manually request legacy grub
[04:01] bgoldsmith: it should be about the same as grub legacy + raid1
[04:02] Fair enough. I'm off to bed, anyway.
[04:02] the only thing that I need to be able to do is fool grub into believing that it is installing on (hd0) when it is installing on sdb.
[04:03] can grub legacy deal with the boot partition on LVM + RAID ?
[04:03] twb: i have successfuly configured bind9 i forgot to edit one of the A name section from ns to ns1
[04:03] jongbergs: OK
[04:03] twb: thanks for your help
[04:04] twb: is there a default log for bind9?
[04:04] jongbergs: buggered if I know
[04:04] jongbergs: ask /var/log/
[04:04] making grub believe that it is installing on (hd0) when installing on /dev/sdb can be done with device (hd0) /dev/sdb and root (hd0,0)
[04:04] jongbergs, bind logs to /var/log/syslog IIRC
[04:05] a grep -R should find it
[04:05] what is the official way of doing that with grub2 ?
[04:05] yay i fixed my server issue by doing ufw allow 80
[04:05] qman__: i mean for bind9 log?
[04:06] jongbergs, so do i
[04:07] twb: seems that no log facility for bind9, i can't find it at /var/log
[04:07] it doesn't have its own log by default
[04:07] twb: any idea?
[04:08] qman__: i see..i'll try to check the ubuntu doc to find out..thanks
[04:12] jongbergs, I'm not sure what sort of logging you're after
[04:12] it logs all errors to either /var/log/syslog or /var/log/messages, can't remember which it is
[04:17] Say if i had a 16TB server for file storage. A few months from now I use up all 16TB of space, so i buy another server with 16TB of storage, so now i have 2 16tb servers. What system would i need to put in place so i can combine those to servers into a single 32tb storage space? Would it be a Cluster File System? LVM? what?
[04:18] marks256, anything like that would have to have been planned ahead of time
[04:18] I don't know how specifically to go about it but I know that much
[04:19] qman__, well of course. that's what i'm asking for. i need to know what infrastructure should be set in place
[04:21] I set up a server and I can connect by remote desktop viewer but I can't see the windows when they open. Any ideas
[04:22] baltadt, you're going to have to be more specific than that, but by the looks of it, you're in the wrong support channel--this channel only supports ubuntu server edition, meaning no GUIs
[04:30] i need help with remote desktop viewer any one here
[04:31] baltadt, remote desktop viewer is a GUI application and is not supported here--see #ubuntu instead
[04:34] marks256 - I was thinking about a drbd based SAN
[04:34] http://www.springerlink.com/content/n358433838h81458/
[04:34] on googling
[04:34] bgoldsmith, and what is that exaclty?
[04:35] ok sorry
[04:44] Hi all. I get dumped in my initramfs with an error about not finding my root device (on lvm). I appear to not have any /dev/md* devices (which contain /boot and my lvm pv). Anybody have any ideas on how to fix?
[04:44] marks256: sorry I
[04:44] am kinda occupied right now
[04:45] marks256 could you please google/wikipedia for SAN and DRBD ?
[04:49] bgoldsmith, i did. i'm not sure that's what i need?
[04:50] would the SAN be used with LVM then?
[04:53] whats the best firewall to use?
[04:56] Alright, so I think I figured it out. It appears the arrays were not in mdadm.conf. Apparently I can use the output from `mdadm --examine --scan`
[04:56] X-M4-X: netfilter/iptables is the only real choice.
[04:56] New bug: #498943 in ntp (main) "ntp-doc not current" [Undecided,New] https://launchpad.net/bugs/498943
[04:57] well i dont care what i have to do, my server and clients will have the best security i can offer.
=== nxvl_ is now known as nxvl
[04:58]
=== cmdshftn_ is now known as cmdshftn
[05:07] Adding my arrays to mdadm.conf and update-initramfs fixed it.
[05:40] I have installed Ubuntu 9.10 alongside an existing Windows 7 installation, but I chose not to install Ubuntu's boot loader for various reasons. When I power on my computer, Windows 7 loads, which is how I want it. I want to manually boot the Ubuntu system from the installation disc. Is this possible? Are there options that I can specify in "Boot Options"? Any information would be...
[05:40] ...greatly appreciated.
[05:42] You should've asked to install grub in the MBR
[05:42] It would then have let you pick either OS, and you would only need to change which was the default
[05:49] I am not the only user of the computer and I did not want the installation messing with the MBR. Is it possible to boot it from the installation disc?
[05:51] Yes, but it'll involve an awful lot of typing each time
[05:52] Oh, not from an install DISC.
[05:52] If you had an install KEY, then probably.
[05:53] Well, it it was extlinux -- I suppose stock ubuntu install key media use syslinux, which won't be able to get the kernel and ramdisk from an ext /boot any more than isolinux would
[05:54] I mostly installed it just to play around with it. I last used Linux many years ago (around 1999) and I just wanted to see what all the Ubuntu hype is about. I was also planning on introducing it to my girlfriend if it is user-friendly enough.
[05:54] But, like I said, I didn't want it touching my MBR.
[05:54] cb1147: The Ubuntu boot loader will let you select Windows to be your default boot. Doing it the way you are trying too is just way hard.
[05:55] Is it possible to set the timeout low enough for my girlfriend to not notice it, but still be able to select Ubuntu if I desire?
=== jongberg1 is now known as jongbergs
=== cb1147 is now known as cwinebrinner
[06:09] Yes
[06:10] I usually set mine to three seconds
[06:10] With extlinux, at least, you can set it to have NO delay, but to open the menu if the Scroll or Caps Lock is on, or if the Shift key is held down.
[06:11] Kinda like the OF on the new world macs.
=== SultansElephant is now known as Gangsta
=== Gangsta is now known as SultanElefante
[07:00] I have this in my cron; @daily /usr/bin/sudo -u /usr/sbin/logcheck logcheck; I assume it runs 0 0 * * *; but at midnight; no email
[07:00] oh....... nevermind /me facepalms
[07:10] @daily is documented in the crontab manpage
[07:10] You also got sudo's args backwards
[07:11] If I were you, I wouldn't give the full path -- instead customize PATH at the top of the crontab if you really must
[07:21] MTecknology: More to the point, why not just edit /etc/cron.d/logcheck instead of creating new (incorrect) cron entries by hand?
=== mylogic_ is now known as mylogic
[07:32] jmarsden: I could do that too...
[07:40] Hi, I've setup samba as a PDC with the default tdb as a user backend. I'm able to join the domain and logon using Windows clients, but I can't do it with Ubuntu clients. I've installed and configured (to the best of my ability) winbind, what else do I need? Do I need Kerberos?
[07:47] logcheck already runs daily, doesn't it?
[07:47] jmarsden: I assumed he was editing that file already :-)
[07:47] twb: Hourly plus at reboot, by default ...
[07:47] Oh, I'm confusing it with logwatch
[07:48] alkisg: I'm not at expert on that, but take a look at the likewise-open5* packages
[07:48] And here, for STUPID reasons logcheck is running every ten minutes :-/
[07:48] jmarsden: thank you, /me looks...
[08:13] How does ufw merge the .rules files?
[08:13] It can't be just calling iptables-restore on each one in turn...
[08:14] Oh, maybe it's using iptables-restore --noflush
[09:00] morning,
[09:01] regarding DKIM and multiple domains - I have many domains which use the mail server 'mail.example.com', so do I need to get DKIM to sign for 'mail.example.com' or 'client-domain.com'?
=== |eagles0513875| is now known as eagles0513875
[09:06] AlexC_: presumably you can work that out by attempting to validate a test email that comes out of it?
[09:11] Bloody hell, do you reckon gufw has enough dependencies?
[09:14] Through gksu, it manages to pull in udev and devicekit
[09:51] Hi, quick question which google doesn't seem to provide a quick answer for. I'm getting an error when I try and start bind after an upgrade to Karmic, where do I find its logs? (or make it verbose?)
[09:52] elliotjhug: for the latter you presumably want /etc/named/named.conf or something
[09:52] (Just a guess)
[09:53] twb, logs would be preferable to changing even more settings (since its likely a config error causing my trouble)
=== alex88[sleep] is now known as alex88
[10:23] mathiaz: apropos https://blueprints.launchpad.net/ubuntu/+spec/server-lucid-puppet-etckeeper-integration
[10:24] mathiaz: There are a few unrealistic expectations there
[10:24] "implement proper file permissions and ownership support in bzr" is quite complex
[10:25] "etckeeper: fix .bazaar/ owned as root bug" is also difficult
[10:26] "implement new features/commands" is doable, but if we want the feature to go back upstream, we'll need to do it in a VCS-agnostic way
[10:27] ttx: ok - for the 1st point, I gues this is taken care of by etckeeper
[10:27] mathiaz: hm, I'm not sure I follow you, then
[10:28] ttx: the current design solution we've discussed with puppet upstream is to use a pre/post commit hook
[10:28] ttx: where puppet would just call etckeeper before/after its run
[10:28] right
[10:28] ttx: the same as the apt integration is done
[10:28] ttx: etckeeper doesn't keep track of ownership/permissions ?
[10:28] mathiaz: ah, yes it does. Not in the best possible way, but it does
[10:29] (the best possible way being using a VCS that is natively aware of those)
[10:30] ttx: right
[10:30] when you said "implement proper file permissions and ownership support in bzr" I thought you meant patching bzr to natively support those
[10:30] ttx: AFAIK no VCS support taht
[10:30] ttx: correct
[10:30] i was having some input output errors with my hdd, now I can't even detect it as a device to partition it, is the hdd just dead now?
[10:30] mathiaz: that's what I found unrealistic, and not necessary to the completion of the spec
[10:31] ttx: probably - I was mainly recording the different points of the plan
[10:31] ttx: we should at least engaged with the bzr team to see what would be required to have ownership/permissions properly handled by bzr
[10:32] mathiaz: I engaged that discussion with them already
[10:33] mathiaz: they recommended to keep it out of bzr and implement it as a plugin that would integrate the .etckeeper metadata
[10:33] mathiaz: rather than a VCS-native implementation
[10:33] basically they don't see the need for native permissions/ownership support in the VCS, just to cater for a specific VCS use
[10:34] ttx: right - I guess that meta-data versioning is not that important
[10:34] mathiaz: the problem is, the non-native approach is bug-prone
[10:34] ttx: using a plugin would work - as long as relevant bzr commands handle permission/ownership correctly
[10:34] mathiaz: that's the thing, the plugin has to modify all those commands
[10:35] see comments on https://bugs.launchpad.net/ubuntu/+source/etckeeper/+bug/322327
[10:35] Launchpad bug 322327 in etckeeper "Integrated permissions/ownership diff output for etckeeper/bzr" [Wishlist,Confirmed]
[10:35] and https://bugs.launchpad.net/ubuntu/+source/etckeeper/+bug/322339
[10:35] Launchpad bug 322339 in etckeeper "Integrated permissions/ownership reset on file operations for etckeeper/bzr" [Wishlist,Confirmed]
[10:35] Diff is relatively easy
[10:35] since it's quite limited in scope
[10:36] but getting the permissions/ownership correct on all file operations...
[10:37] mathiaz: anyway, those are etckeeper usability improvements, not /necessary/ to have puppet integration
[10:38] ttx: right
[10:38] ttx: the specification covers both puppet and etckeeper
[10:38] the .bazaar issue is more pertinent
[10:40] mathiaz: maybe we should split it to have a realistic lucid target
[10:40] or it will just be deferred, methink
[10:40] ttx: agreed - the current whiteboard state is more of a dumping ground for high level tasks that need to be done
[10:40] on the etckeeper side I'd keep "in main" and ".bazaar"
[10:41] ttx: the actual plan and work items need to be refined
[10:41] mathiaz: I understand, it's just that we tried to review the status of the spec last Friday with jos
[10:41] * mathiaz nods
[10:41] to see if it was pertinent for alpha3+
[10:42] ttx: ok - so what I could do is to update the whiteboard with a reasonable list of work items for lucid
[10:42] ttx: but I'd rather not loose the other things
[10:42] agreed. You can has a "Future work" section in the whiteboard
[10:42] have, even
[10:42] ttx: the more high level/blue sky items (such has proper bzr integration)
[10:42] ttx: great - I'll update the whiteboard now
[10:43] cool, thx
[10:44] ah, the very fresh daily ISo now has an installable UEC 1.6.2
[10:44] * ttx rsyncs
[10:45] ttx: whiteboard updated - let me know if it seems a reasonable target for lucid now
[10:45] mathiaz: sure, looks ok for me
[11:12] hi there. i installed a new hardisc in ubuntu server 8.10, create a partition and mkfs.ext3 the new partition. Now i want to mount it via fstab, "blkid /dev/sda1" tell me the uuid, and i want to make an entry in fstab with this uuid, but there is no symlink in /dev/disk/by-uuid for that uuid. Do i need to restart a service? Or should i create it manually? Who creates this symlinks in /dev/disk/by-uuid? THANKS
[11:34] I want to use uuid to mount a partition via fstab. Should the symlinks in /dev/by-uuid to /dev/sdXX created automaticly when i create/format a partition? Or do i need to create them manually? Cause my new partition is not listed there
[11:36] did you mkfs.ext3?
[12:32] hi guys, is there any reason why dansguardian is not built with --enable-clamd in hardy security?
[12:46] New bug: #499029 in eucalyptus (main) "[lucid] eucalyptus-nc upstart script fails to start" [High,Triaged] https://launchpad.net/bugs/499029
[12:53] arj: i do mkfs.ext3, but have no symlink in /dev/by-uuid
[12:55] foolano: maybe because its in universe and probably synced from debian? :)
[12:59] foolano: If it's not, it's a mistake. Please file a bug.
[12:59] zul: No, both Debian and Ubuntu use clamav.
[13:00] cemc: ^^^ Could you have a look?
[13:02] IIRC, dansguardian in karmic is built with enable-clamd
[13:07] what will be the best way to virtualize many physical machine into one?
[13:41] foolano: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=533024. as far as i can tell, it was disabled, then re-enabled in dansguardian 2.10 ?
[13:41] Debian bug 533024 in dansguardian "dansguardian: no clamd support" [Important,Fixed]
[13:44] cemc: We should look into making sure we have it enabled.
[13:45] ScottK: alright. There is clamav support, just not clamd
[13:45] Ah, I see.
[13:45] cemc: I'd check hardy-backports and make sure it's enabled there.
[13:45] but I guess we synced that one from debian and it got disabled there
[13:45] nope, it isn't in backports either
[13:46] We're going to do backports -> security soon, so no point in fixing it security if it's OK in backports
[13:46] Probably more than just Hardy then
[13:46] ScottK: in backports it's the same package than in security only with some libclamav6 transition stuff
[13:46] OK.
[13:47] in 2.10 however in Karmic it IS enabled
[13:51] We should update the backports to match that then.
=== freeflyi1g is now known as freeflying
[13:57] when I do apt-get install apache2; It installes the mpm-worker. how do I change it to prefork?
[13:59] a_ok: try apt-get install apache2-mpm-prefork
[14:00] mathiaz: rest are deps of course... why didn't I think of that?
[14:00] thanks
[14:03] ttx: hi
[14:04] mathiaz: o/ again
[14:04] ttx: I've updated https://wiki.ubuntu.com/ServerLucidCloudConfig with a sample config.yaml file
[14:04] ttx: I'd like to get request on the proposal
[14:04] request ?
[14:04] ttx: should I send to ubuntu-ec2@? or is there another mailing for that?
[14:04] ttx: I'd like to get feedback on the proposal
[14:04] ah
[14:05] mathiaz: I'm not up to date on cloud lists
[14:05] ttx: to which mailing list should I send the request for feedback
[14:05] ttx: ok
[14:05] ttx: I'll email the world then
[14:05] mathiaz: just a sec
[14:05] mathiaz: smoser is working today
[14:06] mathiaz: he should know
[14:06] ttx: right
[14:06] ttx: I'll track him down
[14:06] hello all
[14:07] kirkland`, ping, bug 498174 . do we still really need to create static dev entries ?
[14:07] Launchpad bug 498174 in eucalyptus "Maximum number of 8 loopback devices low for SC" [High,Fix released] https://launchpad.net/bugs/498174
[14:08] smoser: hm - here you are!
[14:08] uhoh
[14:08] smoser: I've updated https://wiki.ubuntu.com/ServerLucidCloudConfig with a sample config.yaml file
[14:09] smoser: I'd like to get feedback on the proposal
[14:09] smoser: to which mailing list should I send the request for feedback?
[14:09] smoser: ubuntu-server@? ubuntu-cloud? ubuntu-ec2?
[14:11] what does | do ?
[14:12] smoser: http://en.wikipedia.org/wiki/YAML#Block_literals
[14:12] mathiaz: question for you
[14:12] http://bazaar.launchpad.net/~ubuntu-core-dev/eucalyptus/ubuntu/annotate/760/debian/eucalyptus-common.postinst
[14:12] smoser: newlines are preserved
[14:12] mathiaz, sure, you can send to those lists. i think that makes sense.
[14:12] apparently the "chmod 04754 /usr/lib/eucalyptus/euca_rootwrap" never gets run on a UEC fresh install
[14:13] what should be $2 in that case ?
[14:14] ttx: new install $2 = ""
[14:14] so why oh why
[14:14] ttx: is the rest of the code under if [ -z "$2" ]; then fi run?
[14:15] ttx: it may be an issue with if ! dpkg-statoverride --list /usr/lib/eucalyptus/euca_rootwrap; then
[14:15] mathiaz: I'm not sure
[14:15] there isn't any dpkg-override in the fresh install
[14:16] ttx: right - dpkg-statoverride may be buggy - or the call has changed
[14:16] looks like everything else is alright, but maybe it gets picked up by some other pre-flight check
[14:16] returns the right thing for me
[14:17] * ttx enters deep testing mode
[14:17] ttx: is something else changing the permission on euca_rootwrap (init script)?
[14:17] mathiaz: I like that idea, but no, not that I can find
[14:17] * mathiaz grabs some lunch in the meantime
[14:23] mathiaz, i dont think {{{ are required, are they ?
[14:23] mathiaz, where are you (physically? lunch? its 9:23 eastern)
=== sgsax_ is now known as sgsax
[14:41] can someone help me fix rubygems?
[14:42] gem install rake gives me a not found error... I find that rather hard to beleave
[14:49] i'm subscribed to a bug [#454898] in karmic. recently, i got an email saying that the bug's status has changed from 'in progress' to 'fix comitted'. if i download the karmic iso today, will the bug be fixed in that copy?
[14:50] StrangeCharm, "fix committed" doesn't mean that it's fixed in an iso or in updates, just that the fix is in a branch or otherwise "ready" to be published
[14:51] cyphermox, how can i build an iso with the fix in it?
[14:51] smoser: I'm in France for christmas
[14:51] smoser: so it was 3:23 PM when you asked me.
[14:51] smoser: {{{ is part of the wiki markup
[14:51] smoser: so no - they're not required
[14:52] oh. thanks. sorry for being dense.
[14:52] duh, smoser
[14:52] (i saw that in the wiki change notification)
[14:56] StrangeCharm, you'd have to build one from scratch and include the package from karmic-proposed, afaict; but I'm not sure how you actually remaster an iso -- you can check on https://help.ubuntu.com/community/LiveCDCustomization
[14:56] StrangeCharm, it would probably be easier to just make the system boot and update it afterwards...
=== astechgeek is now known as techgeek
[15:00] cyphermox, can't boot the system. the bug is in cryptsetup, preventing the system from booting
[15:01] StrangeCharm, there is a workaround described though, do they work for you?
[15:02] though, someone in #ubuntu told me that it might be possible to remaster the risk
[15:02] *disk
[15:02] yes
[15:03] i don't recall the workaround. if there was one described in the bug, then i must have concluded that either a) i couldn't implement it, or b) it was not applicable to my situation
[15:04] cyphermox, ^
[15:05] though, the livecd might not work for me: i use the server installer disk.
[15:06] StrangeCharm, which partition doesn't mount?
[15:09] !help
[15:09] Please don't ask to ask a question, simply ask the question (all on ONE line and in the channel, so that others can read and follow it easily). If anyone knows the answer they will most likely reply. :-)
[15:09] cyphermox, each partition starting from the thrid. that is: /boot mounts. / mounts. then other disks that contain lvms with subsidiaries of / (like /etc, /var, and /tmp) don't get mounted: another process interrupts the second request for passphrase input
[15:09] !op hi
[15:11] Hi, I'm trying to add a partition to a degraded RAID10 array using mdadm. I'm running "mdadm /dev/md0 --add /dev/sdb3" and it gives the error "/dev/sdb3 not large enough to join array" - but it *is* large enough - it's exactly the same partition layout (and disk model) as the other 3 disks in the array. Am I missing something?
[15:13] StrangeCharm, could you get to recovery mode?
[15:13] yes, absolutely
[15:13] ok, from there I guess you still can't manually mount the partitions?
[15:14] i can
[15:14] ok
[15:14] but i don't want to have to
[15:14] all of them?
=== sommer_ is now known as sommer
[15:15] i can mount all of the disks that i setup in the installer, and boot the system ok from there, and that's how it's running right now
[15:15] what I'm getting at is you could mount them, then update cryptsetup, reboot, and see if the updated packages fixes the problem.
[15:15] haha
[15:15] mathiaz: I solved my mystery
[15:16] ttx: does this involve some smelly cheeses and wine?
[15:16] nah -- chown root:eucalyptus /usr/lib/eucalyptus/euca_rootwrap clears the setuid
[15:16] cyphermox, you're right, i probably could. i was hoping for a solution that would allow me to make a clean install, without any faffing about
[15:16] it's run just after
[15:16] question is, how could it work before
[15:17] StrangeCharm, it's slightly more work, but I would probably do that before remastering the ISO. at least to be sure that the fix really does correct the issue for you
[15:17] ah, was done through dpkg-statoverride
[15:18] well, the computer hasn't need a reboot since i installed 9.10, so maybe i should just wait until 10.4lts, where the issue should certainly be fixed
[15:19] StrangeCharm, according to the bug report the cryptsetup package in lucid does have the fix already
[15:19] however, in my experience, it's best not to rely on hoping a system doesn't need rebooting :)
[15:20] the alpha/beta for lucid?
[15:20] yes
[15:20] well, it's so infrequent, i'm happy to do it by hand on those occasions, until i can make a clean system that i know will work
[15:23] I have backports enabled but from one package I want to install the original version how do I do that?
[15:25] a_ok: sudo apt-get install packagename=specific-version also check out pinning versions
[15:25] genii: thanks
[15:26] a_ok: "specific-version" is the version number as like seen in the result of an apt-cache policy packagename command
[15:27] New bug: #499048 in eucalyptus (main) "[lucid] euca_rootwrap is not installed setuid" [High,In progress] https://launchpad.net/bugs/499048
[15:30] genii: well it's more the result of if one version does not work than perhaps the other. seems the only solution is stay away from apt and build from source
[15:33] a_ok: I'd suggest using the: sudo apt-get install packagename=specific-version way, then whichever works, pin it
[15:33] !pinning
[15:33] pinning is an advanced feature that APT can use to prefer particular packages over others. See https://help.ubuntu.com/community/PinningHowto
[15:34] genii: none work. but thanks for the info on versions and pinning though. will use that in the future
[15:34] smoser: about server-lucid-ec2-config
[15:35] smoser: how about using upstart jobs instead of writing a complete plugin system (where we'll have handle dependencies anyway)?
[15:35] smoser: for example, for the sample configuration file I've created
[15:36] smoser: there would be two upstart jobs: one that configures all the apt configuration using data from config.yaml
[15:36] smoser: and another upstart job to install package once the apt configuration has been done
[15:37] smoser: extending the syntax is just a matter of writing an upstart job
[15:37] regarding dependencies. i dont think you necissarily have to deal with them.
[15:37] smoser: and sticking it into the right position in the boot process
[15:38] well, maybe you do. boot hooks processes in order received.
[15:38] smoser: well - in my example the package installation needs to be run *after* apt repositories are setup
[15:38] so whatever generated that user data woul dhave to deal with dependencies.
[15:39] i agree that writing upstart jobs is a good way to make sure that things run at the right time.
[15:39] but how are yo ugoing to read that config file and write the upstart jobs?
[15:39] smoser: and you can plug them at any point during the boot process
[15:40] smoser: I thought that the boot hooks would do a dump of the user-data file
[15:40] smoser: or rather of 169.XXX
[15:40] the thing that reads the config file and writes the upstart job stilli needs to be done, and has to have logic for how to handle eahc of those config stanzas, right ? unless i'm missing something.
[15:40] smoser: which should contain the user data *date*
[15:40] smoser: well - the syntax is YAML
[15:40] the input to the "config" is a config section like you showed
[15:41] boot hooks would take care to feed that config to *some* program
[15:41] smoser: how about the following workflow:
[15:41] smoser: 1. boot hooks put the user-data yaml file somewhere (ex : /etc/instance_config/config.yaml)
[15:41] smoser: 2. apt_setup upstart job runs
[15:42] smoser: and parses the the config.yaml fie
[15:42] smoser: so we need to write an upstart job that does the right thing with config.yaml
[15:42] smoser: boot hooks don't call anything - we rely on upstart to do that
[15:42] smoser: it could generate an upstart event
[15:43] smoser: the point here is to leverage upstart dependency solver
[15:43] smoser: and one can hook anywhere into the boot process
[15:44] smoser: so upstart jobs would have to be written to support existing configuration data in config.yaml (like a apt conf upstart job, a package install upstart job)
[15:45] i'm not opposed to the above.
[15:45] smoser: but there is no need to write a framework that handles plugins and calls them accordingly.
[15:45] smoser: it seems to be the easiest way to implement what we want to me.
[15:45] then to insert a handler ("plugin") boothooks would need to be able to wirte an upstart job, and that upstart job would just read the config from some locaiton.
[15:46] what i dislike of the above is static entries in /etc/init that do nothing
[15:47] unless i'm missing something
[15:47] smoser: how about a handler boothooks that just reads the user-data option and dumps it somewhere?
[15:47] smoser: well - you're right. there would be upstart jobs in /etc/init/ that would not be used if there wasn't any config user-data
[15:47] i'm ok with boothooks writing user data to a well known location
[15:48] smoser: right - and then all of other upstart jobs could check wether there is a config.yaml file in the user data
[15:48] smoser: if not, they don't do anything
[15:48] so, there would be a a .conf file in /etc/init for each "type" of supported config
[15:49] smoser: yes - there would be an /etc/init/apt_conf.conf
[15:49] so if you support 75 different config sections, you've got 75 different files in init
[15:49] smoser: and /etc/init/packages_install.conf
[15:49] wait
[15:50] i *think* that upstart will read /etc/init/*/.conf (recursively find .conf files in /etc/init)
[15:50] smoser: there isn't a 1:1 mapping between section in the config.yaml file and plugins (ie /etc/init/*conf)
[15:50] thats my ony real issue in the above, is a bunch of jobs sitting there doing nothing
[15:51] smoser: upstart jobs are free to use whathever part of the configuration files they find useful.
[15:51] no, there isn't a 1:1, i understand that, but whatever have, where you would want to break them up
[15:51] then 1 per *that*
[15:51] which could be lots
[15:51] smoser: agreed.
[15:51] smoser: another goal of the spec is to *only* support common operation
[15:52] smoser: and to help bootstrap an instance into an existing configuration management infrastructure
[15:52] smoser: I'm not convinced we would end up with lots of additional upstart jobs
[15:52] fair
[15:55] i'd like to request that upstart jobs at least go into /etc/init//*.conf. and we should verify the costs of inactive static jobs .
[15:56] my guess is it is small, but at very least for any task that checks for existance of a file (user data) you're goign to spawn a subprocess.
[15:56] i think we're probably not that anal about boot at this point, but it is wasteful
[15:56] smoser: right.
=== robbiew is now known as robbiew_
[15:57] smoser: we could an upstart event to say whether there is a config.yaml file available in user data
[15:57] smoser: that way all relevant upstart jobs could depend on a user-data-config upstart event.
[15:58] smoser: we could *user* an upstart event
[15:58] true. start on (ec2-config-yaml and whatever else)
[15:58] right. and have boothooks emit that
[15:58] i do like that mathiaz
[15:59] mathiaz, how do you want to handle multiple configs
[15:59] you want htem all in one file then?
[15:59] my plan was to call the configurator each time there was a 'part' that was a config section
[15:59] smoser: what do you mean by multiple configs?
[16:00] user data is multi-part
[16:00] smoser: right - so the config.yaml would be a mime part
[16:00] there could be multiple parts that are config type
[16:00] smoser: ok - just concatenate them
[16:00] smoser: IIRC you can cat multiple yaml file
[16:00] smoser: and it will be a valid yaml document
[16:01] yeah. thats fine.
[16:02] smoser: could we skip the mimepart document and stick a config.yaml directly in user-data?
[16:02] smoser: the same way as scripts are handle for now
[16:03] smoser: that way one could actually do things like: --user-data "packages:\n - puppet"
[16:03] smoser: or --user-data "packages: [puppet, postfix ]"
[16:03] mathiaz, yes, thats the goal
[16:04] user data is one of 4 types
[16:04] a.) #! script -- its run at S99
[16:04] b.) parseable yaml.conf possibly passing some other test to make sure that we dont mis-fire
[16:04] c.) mime-multipart message
[16:05] d.) unknown - do nothing
[16:05] make sense?
[16:05] smoser: yes!
[16:05] thats why its important that the config be human typeable
[16:06] smoser: right - I've looked at YAML and came up with the config file I've published
[16:06] smoser: I've also looked into ini syntax - but it would require some tricks to be able to express the kind of structure I've put in the example
[16:07] smoser: YAML seems very natural for that task IMO
[16:08] Human readability is a major design goal for YAML.
[16:09] smoser: how about using user-data-yaml-config as the upstart event that the boothook would fire?
[16:09] smoser: and that other upstart jobs (like apt_conf) would depend on (start on user-data-yaml-config)?
[16:10] right
[16:10] that is fine for a name, mathiaz
[16:10] one other thing i'd like to add to it
[16:10] is that on first boot boothooks will emit user-data-yaml-config-firstboot
[16:11] smoser: hm - I though we would only cover firstboot
[16:11] well, how would you *not* fire every time then?
[16:12] smoser: apt configuration, package installation
[16:12] smoser: landscape registration, puppet registration
[16:12] generic
[16:12] smoser: you'd wanna do that only on first boot
[16:12] smoser: not every time the instance reboots
[16:12] it is quote possilbe that something wants to run based on that config every time
[16:13] smoser: right - one use case I haven't factored in is ebs
[16:13] in ec2-init, in karmic, there are two types of "how often to run"
[16:13] "once ever"
[16:13] and
[16:13] well, actually 3
[16:13] a.) once-ever
[16:13] b.) once-per-ami
[16:13] smoser: that being said we could just emit an upstart event on first boot as wel
[16:13] c.) maybe only 2
[16:14] mathiaz, yes. but it is possible that some config thing would *want* to run every boot
[16:14] smoser: and then package_install would start on first-boot and user-data-yaml-config
[16:14] so i'd fire 2 events
[16:14] right
[16:14] and there is also a 'once-ever' event
[16:14] smoser: the user data doesn't change between reboots?
[16:15] it does not change between reboots
[16:15] i'll tryo to give an example
[16:15] (and try to type)
[16:15] a.) once-ever
[16:16] right now, localization (setting of default-locale) and apt mirror is only run once ever. ie, on the first boot. after a rebundle it will not run again.
[16:16] b.) once-per-ami
[16:17] generate of ssh keys is done once per instance, which in ec2 turns out to be "first boot" because for any given thing that can read that filesystem first boot == instance
[16:17] c.) it doesn't seem at all unreasonable to me that someone wants to writei a plugin that ccan parse user-data config syntax but wants it to run every boot
[16:18] maybe "call-home-on-boot: http://foo.bar.com/wark/?id=akdkdhjdyek"
[16:18] * mathiaz nods
[16:19] mathiaz, i like the upstart jobs. thank you for that idea.
[16:19] smoser: for the last use case, wouldn't it make sense to leverage config.yaml?
[16:19] smoser: in the sense that one would write an call-home-one-boot upstart job
[16:19] smoser: and then start an instance with --user-data "call-home-on-boot: http://foo.bar.com/wark/?id=akdkdhjdyek" ?
[16:20] yes. absolutely.
[16:20] they would
[16:20] or i would want them to
[16:20] but that would then have:
[16:21] start on (local-filesystems and user-data-yaml-config and user-data-yaml-config-boot)
[16:21] or what not. ie, some way that it can be called on *every* boot, but the other (apt config and such) run only on first
[16:22] smoser: right.
[16:23] smoser: I wouldn't call it user-data-yaml-config-boot - first-boot or something shorter
[16:23] smoser: it's not really related to user-data-yaml-conig
[16:23] this is true. good point.
[16:23] smoser: more related to the instance booting itself.
[16:24] * smoser gives mathiaz a high five. thank you for makign this better.
[16:24] smoser: that means you'd have 3 upstart event: first-boot-ever, first-boot, reboot
[16:24] smoser: and may be prefix them with cloud-
[16:25] smoser: and I don't think you need to have cloud-reboot
[16:25] smoser: as jobs using that event would just ... well ... run at every boot
[16:26] smoser: which is the default case for upstart jobs
[16:28] smoser: I'd suggest to rename user-data-yaml-config to cloud-yaml-config - to make it a little bit more generic
[16:28] smoser: and the last bit for the server-lucid-ec2-config spec would be to settle on a local filename
[16:29] smoser: where the config.yaml would be stored
[16:30] mathiaz, s/cloud-yaml-config/cloud-config-yaml/ ?
[16:31] smoser: the non-double64-challenged eucalyptus landed in karmic-updates, yay
[16:31] smoser: hm - I wonder if we really need the yaml part
[16:31] smoser: cloud-config
[16:31] non-double-base64-challenged, to be more exact
[16:32] first mover claims main namespace, i'm good with that
[16:32] mathiaz, regarding location, right now things are ec2
[16:32] smoser: the question is what happens if we away from YAML later?
[16:33] so /var/lib/ec2 is where it woudl "fit"
[16:33] err... but i think maye this is more run-time, so i think
[16:33] smoser: right - considering that we're running on UEC (and potentially other clouds) I'd move way from ec2
[16:33] smoser: that's I why I suggest cloud
[16:33] /var/run//ec2/config.yaml
[16:33] smoser: as the prefix
[16:33] i agree with the cloud name
[16:33] /var/run/cloud/config.yaml
[16:34] err, s/ec2///
[16:34] i like that.
[16:34] smoser: is there any reasons to specify that you're running on EC2 rather than UEC?
[16:34] not for this, no.
[16:35] but it is quite reasonable that a job would need to know that.
[16:35] so it will be availble to jobs
[16:35] smoser: agreed.
[16:36] smoser: if we provide /var/run/cloud/config.yaml as the standard location for instance configuration - how easy is it to support other Clouds?
[16:36] smoser: IIRC the rackspace cloud doesn't use the user-data option
[16:36] they do have per-instance data (i believe)
[16:36] smoser: my point here is to try to make config.yaml cloud agnostic so that we don't have to rewrite the upstart jobs
[16:37] smoser: right - the mechanism for making it available is different than in EC2/UEC though.
[16:37] and if the end user has the abillity to insert customized per-instance data, then we're good. it is boot hooks job to copy config-yaml instance data to that location
[16:37] smoser: right
[16:37] i agreed with your point up there. yes.
[16:38] one thing....i kind of dont like...
[16:39] smoser: ok - to summarize: boothooks will provide a cloud config.yaml and then emit an upstart even cloud-config with CFGFILE=/path/to/config.yaml
[16:40] smoser: and -ec2-config upstart jobs will start on cloud-config and use CFGFILE to get the configuration file
[16:40] it need to be fixed in upstart.
[16:40] to insert a config handler into an existing instance, the user will then:
[16:40] a.) provide config data
[16:40] b.) provide a mime-part that is recognized by boothooks as a "config handler" and boothooks will then put it into /etc/init/cloud/ with a given nanme
[16:40] the thing that stinks is
[16:40] mathiaz, yes.
[16:40] now, continuing above.
[16:41] the thing that stinks about a and b above is that if they wish to write their config handler in any thing othe rthan posix shell , then they have to
[16:41] ttx: will your recent changes in auto detecting the cloud controller make it to tomorrow's image?
[16:41] i need to set RAID0 in ubuntu server 910.help me.thanks
[16:41] c.) provide a mime-part that is stored in a file and exectued from b.)
[16:41] because 'script' can only posix sh
[16:42] TeTeT: yes it should
[16:42] We should have an intermediary build before that, though
[16:43] TeTeT: the workaround is quite easy, if you need to get passed that
[16:43] smoser: right.
[16:43] smoser: at least there is a workaround for now
[16:44] smoser: and boothooks could probably automatically generate the script part for the upstart job
[16:44] smoser: so that the end user would have to generate two mime-parts:
[16:44] smoser: 1. an upstart job with the correct dependency line
[16:45] smoser: 2. the actuall handler
[16:45] TeTeT: just replace "avahi-publish -s $CLOUD_IP_ADDR" by "avahi-publish -s CLC" in /etc/init/eucalyptus-cloud-publication.conf
[16:45] that will avoid the conflict and let the announce succeed
[16:45] (then sudo stop eucalyptus / sudo start eucalyptus)
[16:45] smoser: then the boothook would actually copy the 2nd part to a known place and add the necessary script section to the upstart job
[16:46] well, yeah, but there may be other 'start on' pieces and such
[16:46] its not that bad.
[16:46] ttx: thanks! giving the ip address during the install worked fine for me as a work around. Want to test the split of CLC and CC later this week again. didn't work for me last Thursday/
[16:46] heres how you can do it in a #!/bin/sh -e 'script' section
[16:47] TeTeT: at that point I only test the "old world"
[16:47] mathiaz, http://paste.ubuntu.com/344173/
[16:48] smoser: right
[16:48] smoser: so you could generalize that to: exec /path/to/new/handler
[16:48] :)
[16:49] smoser: that would require to have a well-formed script
[16:49] smoser: which is reasonable IMO
[16:49] ttx: ok, I volunteer to try the new stuff ;)
[16:50] TeTeT: I have yet to look at /what/ the installer is supposed to do
[16:50] TeTeT: like, only sync the keys, or download a preseed
[16:50] and in the latter case, what's in that preseed :)
[16:51] but at this point I'm trying to /not break/ the classic topology
[16:51] ...when introducing the crazy sauce.
[16:52] TeTeT: given that 1.6.2 was pretty broken, I'm busy making it work too :)
=== robbiew_ is now known as robbiew
[16:53] a well formed sscript, mathiaz ?
[16:53] what?
[16:54] smoser: hm - nevermind
=== robbiew is now known as robbiew_
[16:58] smoser: I've updated https://wiki.ubuntu.com/ServerLucidCloudConfig with what we've discussed above
=== dendro-afk is now known as dendrobates
[17:03] thanks mathiaz
[17:20] Hello, I'd like to test UEC (see my post http://ubuntuforums.org/showpost.php?p=8520901&postcount=9), but got stuck on the automated install. At this stage, can anyone assist, or am I inclined to work through the problem myself.
[17:22] I'm currently configuring a DNS server for my server. The guide i'm following tells me to replace ns1 with the DNS server name. Is that the DNS for my ISP or for my server.
[17:25] xtjacob: configuration of DNS is nothing a beginner should do (except strictly for the local net - for testing purposes only)
[17:26] then is there anyway i can get my website with a hostname without paying or using a free online service?
[17:27] anyone know the command for extracting a rar file unrar is installed already
[17:27] xtjacob: for your own testing or for general availabilty?
[17:27] ls
[17:27] hah
[17:28] * kees tries again...
[17:28] X-M4-X: did you look at man rar?
[17:28] general availability
[17:28] what is that
[17:28] is update-notifier-common installed by default on server installs?
[17:28] oh nvm - no i did not
[17:30] guntbert: general availability
[17:30] !info update-notifier-common
[17:30] update-notifier-common (source: update-notifier): Files shared between update-notifier and adept. In component main, is optional. Version 0.90 (karmic), package size 22 kB, installed size 312 kB
[17:31] xtjacob: then you will have to register a domain anyway
[17:33] gunbert: is there any way to do this for free?
[17:33] New bug: #499128 in samba (main) "package samba 2:3.4.0-3ubuntu5.1 failed to install/upgrade: subprocess installed post-removal script returned error exit status 1" [Undecided,New] https://launchpad.net/bugs/499128
[17:35] xtjacob: yes, several, for instance with no-ip.com or with dydns.org - though they have different targets - but google for "free domain"
[17:36] New bug: #499142 in samba (main) "Dependencies of smbclient remove kubuntu-desktop and prevent its reinstallation" [Undecided,New] https://launchpad.net/bugs/499142
[17:36] i wish i knew who told me to get cpanel on my server - ubuntu isnt a supported OS
=== erichammond1 is now known as erichammond
=== luis__lopez is now known as luis_lopez
[18:20] this may seem like a dumb question, but what do i type in to FTP into my server? I keep getting connection refused
[18:21] did you install an ftp server?
[18:22] i think so how can i tell?
[18:22] dpkg -l|grep ftp
[18:22] and what does netstat -anp|grep :21 say
[18:24] dpkg -l|grep ftp says ii ftp 0.17-19 The FTP client, and netstat -anp|grep :21 says (No info could be read for "-p":geteuid()=100 but you should be root)
[18:25] sudo netstat -anp|grep :21
[18:25] but it looks like you dont have an ftp server install
[18:25] nothing came up
[18:25] apt-get install proftpd/pure-ftpd/vsftpd
[18:26] one of those
[18:26] then do i log in with my username?
[18:26] ok it worked thanks
[18:31] also how do i figure out the external IP for my server?
[18:31] ifconfig
[18:32] which ip do i look at?
=== dendrobates is now known as dendro-afk
[18:49] Any ideas how I can have my backup server pull backups from a windows system?
=== dendro-afk is now known as dendrobates
[18:49] I'd prefer not using samba..
[18:51] I also need to figure out how to have the windows system sent its logs to the logging server
[18:54] MTecknology: i thinkk rsync has cient for win, also possible solutions re ftp server or sftp
[18:54] MTecknology: I haven't done this for a while now since I don't use Win boxes much anymore. But you can install rsync onto Win boxes using cygwin
[18:55] I'll try that out
[18:55] thanks
[18:55] any ideas about having it send logs to another system?
[18:56] brb
=== chmac is now known as chmac_away
[20:28] I'm grying to build libtorrent
[20:28] but when I do ./configure I get an error: checking for OPENSSL... configure: error: Package requirements (openssl) were not met:
[20:28] but i Do have installed openssl
[20:30] hello to all. anybody here that have knoweledge and experience with postfix ?
[20:30] !ask | xperia
[20:30] xperia: Please don't ask to ask a question, simply ask the question (all on ONE line and in the channel, so that others can read and follow it easily). If anyone knows the answer they will most likely reply. :-)
[20:31] frojnd: apt-get build-dep libtorrent
[20:31] or whatever the source package for libtorrent is
[20:31] frojnd: you probably are lacking (at least) libssl-dev
[20:32] Hello
[20:32] lamont: how do you know libssl-dev is the one that is lacking? :( I never build before myself I don't really understand what packages is missing
[20:32] frojnd: because that's where openssl delivers its development files
[20:32] Can someone help me with an issue I'm having?
[20:33] lamont: ok
[20:33] frojnd: apt-get build-dep $package will install the build-deps for that package
[20:33] binhtran: without knowing the issue, how can we possibly answer that?
[20:33] (don't ask to ask a question, just ask)
[20:34] i have my server running, and I put it in my router's DMZ for a while to see if I could get outside access, and now all my links reroute to cheapnames.net
[20:34] okay. i need to teach postfix to send mails with smtpauths. i have changed allready the relayhost setting from "relayhost = [mail.myisp.com]" Port 25 that worked till yet great to "relayhost = [smtpauths.myisp.com]" Port 465.
[20:34] For some reason however postfix still send the email to port 25 and not 465. anybody a solution ?
[20:34] sounds like you have dns issues
[20:34] lamont: and now I typed make and I got an error: ../../../libtool: line 1210: libtool: compile: cannot determine name of library object from `': command not found
[20:34] make[4]: *** [block.lo] Error 1
[20:34] and so on
[20:35] frojnd: debuild -b
[20:35] unless you're trying to package unpackaged source
[20:35] wait, what?
[20:36] frojnd: see also debian/rules (which is a makefile) for exactly how the source wants to be built
[20:36] lamont: debuild: fatal error at line 606:
[20:37] lamont: cannot find readable debian/changelog anywhere! Are you in the source tree?
[20:37] binhtran: not sure I know where to start to help you - and pretty much out of time myself
[20:37] frojnd: so you're starting with upstream source, not the debian source package?
[20:37] lamont: I don't know
[20:38] frojnd: start with apt-get source libtorrrent :-)
[20:38] lamont: no
[20:38] I need those versions
[20:38] that has an upstream source, and a diff between that and the debian package
[20:38] if I wanna to do wtorrent
[20:38] frojnd: ah, then in that case, see the docs inside the source for how to build it.
[20:39] lamont: yeah
[20:39] I'm definetally gonna have to learn how to build this
[20:39] :|
[20:39] frojnd: the debian package will tell you how the debian maintainer decided to do it...
[20:39] but yeah, enjoy
=== Jeeves__ is now known as Jeeves
=== Jeeves is now known as Jeeves_
[21:02] New bug: #499129 in samba (main) "package samba 2:3.4.0-3ubuntu5.1 failed to install/upgrade: subprocess installed post-removal script returned error exit status 1 (dup-of: 349469)" [Undecided,Confirmed] https://launchpad.net/bugs/499129
[21:31] hello
[21:31] I want to run mkdir /home/samba/backups/`date +%Y-%m-%d`
[21:31] on a daily cron
[21:32] but it doesnt work when i do something like 0 1 * * * mkdir /home/samba/backups/`date +%Y-%m-%d`
[21:34] anyone able to help?
[21:35] _{what does it do instead?
[21:36] nothing
[21:36] running the command, makes the folder
[21:36] but in cron
[21:36] it creates nothing
[21:36] _{logging cron anywhere?
[21:37] dont think so
[21:40] you're missing the user to run the job as
[21:40] should be 0 1 * * * root mkdir /home/samba/backups`date +%Y-%m-%d`
[21:41] _{hm. i had assumed it was in a user's cron but i am wrong to assume anything :D
[21:41] true, that wouldn't be necessary in a user crontab
[21:41] but I always use cron.d
[21:41] easier to manage
[21:45] I-Blocklist069: % is a special character in crontabs
[21:47] You can read about this in 'man 5 crontab'
[21:48] so have to comment it out?
[21:48] no, just quote the path
=== chmac_away is now known as chmac
[21:49] i tried to put PATH=/usr/sbin:/usr/bin:/sbin:/bin
[21:49] at the top
[21:49] that's irrelevant
[21:49] 0 1 * * * root mkdir "/home/samba/backups`date +%Y-%m-%d`"
[21:50] like that
[21:50] qman__: no, that is incorrect
[21:50] oh, need to escape manually?
[21:50] You should read the bit in 'man 5 crontab' about % signs, as I said
[21:50] oh yeah
[21:50] forgot, crontab is tricky there
[21:51] is it possible to VNC into a headless server without a gui from my laptop?
[21:52] I can ssh in without issue, but there are still things I am need to do with a gui
[21:52] Xpistos|work, VNC only works when a GUI is running
[21:53] no idea why you'd need one of those stupid things, though ;)
[21:55] qman__: Thanks for the help. I can't convert .mov's to avi and compress them to the size file I want without the gui. I don't know enough of the commands to do it in the command line . . . yet.
[21:57] anyone know a FREE WORKING solution alternative to cpanel with a filemanager ehcp and net2ftp is just gay
[22:08] X-M4-X: Looked into ebox?
[22:11] X-M4-X: ScottK: eBox doesn't have an FTP module. So he will need something else
=== dendrobates is now known as dendro-afk
=== dendro-afk is now known as dendrobates
=== erichammond1 is now known as erichammond
[22:49] someone using fuzzyocr here ?
=== dendrobates is now known as dendro-afk
=== dendro-afk is now known as dendrobates
=== dendrobates is now known as dendro-afk
[23:51] so, I stupidly accepted the default partition setup on my 2TB system. Now I have a 1TB swap partition. I chose "Default with LVM", how can I reclaim the space from /dev/mapper/files-swap_1 and allocate it to /dev/mapper/files-root ?
[23:52] (I'm in parted right now)
[23:52] <--beating his head against the wall right now... (dealing with ubuntu install and black screen)
[23:54] can anyone here give a little advice on how to attack this?
[23:54] donspaulding: it'd probably be easier with the LVM tools directly
[23:55] donspaulding: I guess you don't use your swap right now ? (as in, you can temporarily disable it)
[23:55] stgraber: yeah, the machine's got 12GB of RAM, swap can be disabled.
[23:55] ok, so go with: swapoff /dev/mapper/files-swap_1
[23:56] then: lvreduce /dev/mapper/files-swap_1 -L 2G
[23:56] if you want it to be 2GB large afterwards
[23:56] then, you can: mkswap /dev/mapper/files-swap_1 and re-enable it with: swapon /dev/mapper/files-swap_1
[23:57] that should give you a 2GB large swap if all went well
[23:57] if that's the case, then we can resize the root partition to use the free space
[23:57] stgraber: lvreduce gave me….. Volume group mapper doesn't exist
=== dendro-afk is now known as dendrobates
[23:58] is mapper the logical name of a different device I should be passing in to lvreduce?
[23:58] donspaulding: did you include the whole path including files-swap_1 ?
[23:58] like /dev/files/swap_1?
[23:58] stgraber: yes
[23:59] heh
[23:59] quick question for canonical servers what are the naming conventions?
[23:59] donspaulding: ok, then try with: lvreduce files/swap_1 -L 2G
[23:59] I see updates are element names
[23:59] lithium and germanium}}}}