| aljosa | juju.agents.machine using 100% cpu, no lxc instances running. any idea why? | 00:22 |
|---|---|---|
| === Ursinha is now known as Ursinha-zz | ||
| === Ursinha-zz is now known as Ursinha-zzz | ||
| === Adri2000 is now known as Guest74775 | ||
| === Ursinha-zzz is now known as Ursinha | ||
| yaboo | getting the error Credentials cache file '/tmp/krb5cc_1005' not found when I try to do ssh passwordless login from ubuntu 12.04 to centos | 06:52 |
| greppy | yaboo: I think that centos may be looking for kerberos, not sure what you would need to do to make that happy. | 07:28 |
| twb | you get that error in sshd log? | 07:29 |
| yaboo | will look guys | 07:30 |
| blkperl | so what determines which packages the ubuntu-server team is responsbile for? | 07:39 |
| blkperl | i see bacula is on the ubuntu-server teams list, but amanda is not | 07:39 |
| twb | I dunno about the "team", but I answer questions based on whether I care | 07:41 |
| twb | e.g. all the cloud buzzword wank bores me, so I ignore it | 07:41 |
| twb | everything in universe gets basically zero support from canonical, and that is probably the discrepancy you're seeing. | 07:42 |
| blkperl | oh right that makes sense | 07:42 |
| twb | apt-cache policy will tell you where it comes from | 07:43 |
| blkperl | well if your interested in helping me out bug 932064 | 07:44 |
| uvirtbot | Launchpad bug 932064 in amanda "amtape crash" [High,Confirmed] https://launchpad.net/bugs/932064 | 07:44 |
| twb | Not me, sorry | 07:45 |
| cobbler | hi all | 08:21 |
| cobbler | this is the preseed file http://paste.ubuntu.com/1010897/ with precise alternate cd | 08:24 |
| cobbler | imported the alternate cd like this , cobbler import --name=ubuntu-server --path=/mnt --breed=ubuntu | 08:25 |
| cobbler | then added cobbler profile like this, cobbler profile edit --name=ubuntu12_04-x86_64 --kickstart=/var/lib/cobbler/kickstarts/ubuntu-nqa.seed --kopts="auto url=http://192.168.35.1/ubuntu-nqa.seed priority=critical lang=english locale=en_US console-keymaps-at/keymap=us console-setup/ask_detect=false console-setup/layoutcode=us" | 08:25 |
| ikonia | cobbler: this sounds silly, but I don't see a question in there, just you explaining your config | 08:28 |
| twb | Perhaps he's working up to the question | 08:28 |
| cobbler | thanks i went and picked the phone | 08:29 |
| twb | always a mistake | 08:29 |
| cobbler | well the installer ends up prompting me to add a cdrom which is not the case here, trying to automate the install | 08:29 |
| cobbler | do you think my preseed file is not correct? | 08:29 |
| twb | Have you compared your preseed to the one in installation-guide-amd64's appendix? | 08:30 |
| cobbler | this one https://help.ubuntu.com/12.04/installation-guide/example-preseed.txt | 08:31 |
| cobbler | right? | 08:31 |
| twb | I guess so | 08:32 |
| cobbler | i tried reusing the one available here https://help.ubuntu.com/community/Cobbler/Preseed | 08:32 |
| cobbler | and i also read the preseed from the first link | 08:32 |
| twb | I'm not familiar with cobbler, sorry | 08:32 |
| cobbler | np | 08:32 |
| twb | The appendices have several examples of automated preseeding | 08:33 |
| twb | At least they did ten years ago... | 08:35 |
| haxxpop | If I have a port 25 open,how to close it? | 08:53 |
| haxxpop | like this, tcp 0 0 127.0.0.1:25 0.0.0.0:* LISTEN 1421/sendmail: MTA: | 08:53 |
| andol | haxxpop: Well, if you don't want sendmail to listen even on localhost, why do you run it then? | 08:54 |
| haxxpop | I tried to restart exim4 but it notified that port 25 is occupied | 08:56 |
| twb | haxxpop: uninstall your MTA | 08:56 |
| haxxpop | I command "/etc/init.d/exim4 restart" | 08:56 |
| haxxpop | and the output is | 08:56 |
| haxxpop | * Stopping MTA for restart [ OK ] | 08:57 |
| haxxpop | * Restarting MTA [ OK ] | 08:57 |
| haxxpop | * ALERT: exim paniclog /var/log/exim4/paniclog has non-zero size, mail system possibly broken | 08:57 |
| haxxpop | and I open the paniclog, it notified this | 08:58 |
| haxxpop | 2012-05-28 04:41:09 socket bind() to port 25 for address 127.0.0.1 failed: Address already in use: daemon abandoned | 08:58 |
| twb | What does "readlink -f /proc/1421/exe" output? | 08:58 |
| haxxpop | nothing | 08:59 |
| twb | What does "sudo readlink -f /proc/1421/exe" output? | 09:00 |
| haxxpop | I loged in as root. it outputs nothing. | 09:01 |
| twb | Pastebin netstat -nlp output | 09:02 |
| haxxpop | what is Pastebin? | 09:04 |
| twb | !pastebin | 09:05 |
| ubottu | For posting multi-line texts into the channel, please use http://paste.ubuntu.com | To post !screenshots use http://imagebin.org/?page=add | !pastebinit to paste directly from command line | Make sure you give us the URL for your paste - see also the channel topic. | 09:05 |
| haxxpop | http://paste.ubuntu.com/1010943/ | 09:08 |
| twb | If /proc/1421/exe doesn't symlink to something, your system is even more screwed up than I thought | 09:08 |
| twb | It sounded like you'd managed to install both sendmail (as in Allman's) and exim4 at the same time, probably by using some junky horrible thing like Scalix or cpanel, and that they were fighting. | 09:09 |
| twb | But since your /proc is not behaving as expected, you're probably in a VPS or something even sillier | 09:10 |
| twb | You should probably do a clean install. | 09:10 |
| haxxpop | my system is a VPS | 09:10 |
| twb | Then talk to your VPS provider because your system is screwed up | 09:10 |
| haxxpop | Don't exim4 usually work in VPS? | 09:14 |
| twb | That is not a meaningful question. | 09:15 |
| twb | Ask your vendor why there are two MTAs installed. | 09:15 |
| haxxpop | sendmail and exim4 must not be installed at the same time? | 09:16 |
| twb | Correct. | 09:17 |
| twb | It would be like having two engines in your car | 09:17 |
| haxxpop | I got it out. Just remove sendmail and reboot | 09:51 |
| zastaph | how do I find out when aptitude is supposed to upgrade a package? say git from 1.7.5.4 to 1.7.10.x ? | 10:07 |
| zul | jamespage: whats the url for the new bugs again? | 11:56 |
| jamespage | zul: http://reports.qa.ubuntu.com/reports/ubuntu-server/triage-report.html | 11:57 |
| jamespage | zul: time for a quick question? | 12:07 |
| zul | jamespage: noooooooooo :) | 12:07 |
| zul | jamespage: whats up | 12:08 |
| jamespage | zul, OK - so I'm working on an upgrade to solr | 12:08 |
| jamespage | (don't worry - this is not a java question) | 12:08 |
| zul | jamespage: ok | 12:08 |
| jamespage | ATM the solr-tomcat package depends on tomcat6 | 12:08 |
| zul | uh huh | 12:08 |
| jamespage | in my new upgraded package I want to depend on tomcat7 - (tomcat6 is way to old) | 12:08 |
| jamespage | BUT | 12:09 |
| jamespage | when I dist-upgrade from old->new solr-tomcat gets uninstalled | 12:09 |
| jamespage | which was not quite the behaviour I was looking for | 12:09 |
| jamespage | tomcat7 conflicts with tomcat6 | 12:09 |
| zul | so you want something like a transitional package? | 12:09 |
| jamespage | zul: how can I make apt uninstall tomcat6 and install tomcat7 instead? | 12:09 |
| jamespage | if I subsequently do 'apt-get install solr-tomcat' it does that | 12:10 |
| zul | jamespage: erm....too early in the morning i would check on #ubuntu-devel | 12:10 |
| jamespage | zul:ack | 12:10 |
| Daviey | rbasak: can you join #ubuntu-motu please? | 12:49 |
| Daviey | maybe jamespage aswell, based on the changelog | 12:50 |
| SockPants | hi all | 13:15 |
| SockPants | if i want a mail server that only has a number of virtual aliases for different domains and should forward that mail to external email boxes (gmail for instance) do i need anything more than postfix? | 13:17 |
| Soekris | Hello | 13:18 |
| rbasak | SockPants: postfix will be fine for that. | 13:18 |
| SockPants | rbasak: thanks! | 13:18 |
| Soekris | Can some one help me to setup MAAS. i have read the wiki but i'm stuck | 13:18 |
| johnny53287 | hello, i set up a server with ubuntu 11.10 to use the kvm hyperviser and virt-manager. i installed the package "kvm-pxe" to boot the vms via pxe. now when i start a vm, it sends a DHCPDISCOVER, my pxe server response with a DHCPOFFER, but nothing more happens - the vm returns errors like "No IP address". i tested the pxe server with a physical machine and it works well. maybe one of you can give a hint? | 13:22 |
| Soekris | When i want to deploy juju bootstrap . Is it good that in the MAAS console the node have status ready ? | 13:26 |
| hallyn | smoser: open a bug (for launch-index) against cloud-init, or something else? | 13:30 |
| hallyn | feh, ec2 mirrors are sucking today | 14:37 |
| hallyn | utlemming: say, you introduced grub-ipxe, did you ever try to push that to debian? | 15:05 |
| hallyn | utlemming: if not, do you want to try? | 15:05 |
| hallyn | odd, i see the debian bug it closed. but it's not in debian pkg | 15:07 |
| tehgeekmeister | i remember seeing something about a minimal server install, as opposed to the regular server image. does that still exist? where would i find it, if so? | 15:28 |
| qman__ | tehgeekmeister, there is a 'minimal install' option, but that's probably not what you want, you probably want JeOS | 15:30 |
| hallyn | there is the mini iso, which i use for net installs | 15:31 |
| hallyn | http://archive.ubuntu.com/ubuntu/dists/${release}/main/installer-amd64/current/images/netboot/mini.iso | 15:32 |
| hallyn | it's a small download, and if you use a local apt-cacher, you can get pretty low-bw installs too | 15:32 |
| qman__ | if you check the 'minimal install' option during install, it usually breaks your system, as it's literally minimal, lots of stuff you really need not there | 15:32 |
| qman__ | last I checked JeOS was one of the options at boot time | 15:33 |
| tehgeekmeister | yeah, i just want an install with nothing unnecessary on it, and i'll use apt to add anything else on top. i'm not a sysadmin, though, and am just using this for a headless install in a vm on my personal box. | 15:33 |
| qman__ | yeah | 15:34 |
| tehgeekmeister | JeOS seems like it could be right. or the mini iso, worst case, to just skip redownloading outdated packages. | 15:34 |
| qman__ | well, if you're not comfortable, I'd recommend just sticking with the normal system | 15:34 |
| qman__ | it's really not that heavy to begin with | 15:35 |
| qman__ | a base install takes something like 850MB on-disk | 15:35 |
| qman__ | the main reason to use these is to fit into really tight spaces, either for specialized hardware or maximizing performance in virtual environments | 15:36 |
| tehgeekmeister | oh, i'm plenty comfortable doing my own config, i'm just after the smallest functional install to start off of. | 15:36 |
| hallyn | lynxman: smoser: utlemming: I gather you all would prefer we stick with our own ipxe, rebased to most recent git commit, rather than the debian pkg? | 16:13 |
| lynxman | hallyn: indeed | 16:14 |
| hallyn | sigh | 16:14 |
| lynxman | hallyn: sorry to ruin your day :) | 16:14 |
| hallyn | i just would prefer we didn't have to separately support... | 16:17 |
| lynxman | hallyn: I've tried :) | 16:18 |
| hallyn | oh, i see | 16:18 |
| hallyn | thanks | 16:18 |
| hallyn | lynxman: do you feel we need an update from upstream git, or should we stick with waht we have? | 16:19 |
| lynxman | hallyn: we can easily do one with get-orig-source, the package structure is "sane" now so it should be easy | 16:20 |
| hallyn | right, i saw that in the rules (that was my tip-off that you would prefer upstream merge :) | 16:21 |
| hallyn | we can easily do it, but is there any advantage or will we be introducin upstream bugs for no benefit :) | 16:21 |
| hallyn | lynxman: if i put a package in ppa, do you have some tests you can run? | 16:22 |
| lynxman | hallyn: no advantage whatsoever, and yeah I can do some testing but for now it's pretty manual I'm afraid | 16:23 |
| ZenMaster | Good day Gentlemen,. | 16:24 |
| hallyn | lynxman: ok, if no advantage, then nm - i'll just push a package with debian's rom-change-banner-timeout.diff applied | 16:24 |
| hallyn | lynxman: thx | 16:24 |
| lynxman | hallyn: cool :) | 16:24 |
| hallyn | lynxman: did you see that in bug 948323 smb is waiting on you? | 16:30 |
| uvirtbot | Launchpad bug 948323 in ipxe "Rom images for e1000 and ne2k missing vendor and device id" [Low,Confirmed] https://launchpad.net/bugs/948323 | 16:30 |
| jamespage | zul, munin 2.0 from Debian? | 16:31 |
| zul | jamespage: what about it? | 16:31 |
| lynxman | hallyn: ouch... didnt see that one | 16:31 |
| jamespage | zul: I was about to merge it | 16:31 |
| zul | jamespage: be my guest | 16:32 |
| pior | Hello folks! | 16:41 |
| pior | Is there a cloud specific chan ? | 16:41 |
| pior | assuming cloud==AWS | 16:42 |
| utlemming | pior: ##aws is for all AWS questions, but for Ubuntu specific cloud stuff, here is fine | 16:43 |
| pior | utlemming, thanks | 16:43 |
| pior | The AWS apt repository are malfunctionning | 16:44 |
| pior | http://us-east-1.ec2.archive.ubuntu.com/ubuntu/ | 16:44 |
| pior | the ones of US-east-1 at least | 16:44 |
| jamespage | pior: what error are you getting? | 16:45 |
| pior | I get 1/3 of HTTP 200, 1/3 HTTP 403, 1/3 timeout by repeating this : curl -I http://us-east-1.ec2.archive.ubuntu.com/ubuntu/ | 16:45 |
| === Lcawte|Away is now known as Lcawte | ||
| pior | indead there is 3 IPs on us-east-1.ec2.archive.ubuntu.com | 16:46 |
| jamespage | utlemming, anything you are aware of ^^ | 16:46 |
| utlemming | jamespage, pior: not that I am aware of....looking now | 16:47 |
| utlemming | prior: what ubuntu version? | 16:48 |
| utlemming | er, pior: ^ | 16:49 |
| hallyn | lynxman: ok, i have a debdiff ready for the timeout issue, i'll wait a few mins to see if you object to smb's patch. if not i can toss that in as well | 16:51 |
| pior | utlemming, sorry : precise | 16:53 |
| utlemming | pior: looking now | 16:54 |
| pior | utlemming, that's great! | 16:54 |
| lynxman | hallyn: patch looks legit, I'd go for it :) | 16:54 |
| pior | I relaunched the AWS stack (instances) 2 times, it's probably not instance specific | 16:54 |
| utlemming | pior: we have new beta repos that you can use if you want. Run: sed -i "s,ec2.archive.ubuntu.com,ec2.archive.ubuntu.com.s3.amazonaws.com,g" /etc/apt/sources.list | 16:56 |
| hallyn | lynxman: it looks scary to me :) but ok, i'll push it. thanks! | 16:56 |
| utlemming | pior: I'll see if we can't get this fixed ASAP though | 16:57 |
| utlemming | pior: it should be fixed now | 17:09 |
| lynxman | hallyn: all that project is a bit scary ;) | 17:11 |
| hallyn | which, xen, qemu, or ipxe? :) | 17:11 |
| lynxman | hallyn: do I have to choose? lol | 17:11 |
| hallyn | new version pushed, though, we should find out whether it works... as soon as someone runs quantal :) | 17:12 |
| * hallyn looks in zul's direction | 17:12 | |
| zul | works fine for me | 17:12 |
| zul | talking about libvirt right? | 17:13 |
| alibama | hi all - installing ubuntu server and cant get past grub-loader not installing into target | 17:19 |
| alibama | it's my own fault, things were working fine, then i aborted because i hadn't tested the raid config (which turned out to be fine) but now i can't get past this screen | 17:20 |
| hallyn | zul: no, ipxe :) | 17:22 |
| alibama | does anyone have any experience here? i need to manually install grub i believe, however i'm in over my head here | 17:23 |
| zul | oh | 17:23 |
| zul | adam_g: btw your cherrypick for keystone failed on friday | 17:26 |
| hallyn | lynxman: zul: FEH! i dont' have ipxe upload rights. (i was sure i was supposed to) | 17:34 |
| zul | hallyn: should be in the ubuntu-server seed | 17:35 |
| hallyn | lynxman: if you're still around, do you mind pushing http://people.canonical.com/~serge/ipxe-timeout2.debdiff ? | 17:35 |
| hallyn | zul: hm. | 17:35 |
| lynxman | hallyn: I have even less rights than you :) | 17:35 |
| hallyn | zul: do you mind pushing? :) | 17:35 |
| zul | hallyn: sure url | 17:35 |
| hallyn | zul: http://people.canonical.com/~serge/ipxe-timeout2.de | 17:36 |
| hallyn | bdiffhttp://people.canonical.com/~serge/ipxe-timeout2.de | 17:36 |
| hallyn | sigh | 17:36 |
| hallyn | zul: http://people.canonical.com/~serge/ipxe-timeout2.de | 17:36 |
| hallyn | bdiff | 17:36 |
| hallyn | oh | 17:36 |
| hallyn | i see | 17:36 |
| hallyn | (how did that get split in my clipboard? i'm confused) | 17:36 |
| alibama | is this support channel or is there a better place to get some help with my server setup? I've been battling with the grub loader for ~3 hours now | 17:37 |
| hallyn | unless it's a raid/multipath setup, you might just go straight to #grub. but no reason not to ask here since it's server | 17:38 |
| zul | hallyn: done | 17:38 |
| alibama | hallyn: thanks, i'll try grub | 17:38 |
| hallyn | zul: also, were you still going to push the new libvirt, or did you want me to make a new debdiff? | 17:41 |
| zul | did it on friday | 17:41 |
| zul | oh you mean for the thing you forgot | 17:41 |
| zul | debdiff please | 17:41 |
| * zul is stuck in SRU purgatory | 17:42 | |
| hallyn | ok | 17:42 |
| hallyn | i'll just push it then (unless you want to vet it) - thanks | 17:42 |
| alibama | anyone have any suggestions for cleaning up my partitions and getting this server back to ground zero? | 17:43 |
| alibama | i think it's beyond what the grub problems are at this point | 17:43 |
| alibama | ie grub issue is a symptom, not the cause | 17:43 |
| alibama | perhaps someone might know how to test whether i've configured raid correctly? that may be the issue? | 17:48 |
| alibama | do i need to configure this at the beginnng or is it something i can do after the instal | 17:48 |
| alibama | bueller? | 17:49 |
| Soekris | hello I'm stuck with MAAS. I have 1 MAAS server and 2 pxe servers but juju bootstrap don't work about there is no wirteble storage | 17:51 |
| Soekris | But the wiki says nothing about that. the 2 nodes a in status ready | 17:51 |
| Soekris | where must i seek | 17:51 |
| === Lcawte is now known as Lcawte|Away | ||
| pior | utlemming, great! | 18:20 |
| pior | well, 10.250.142.223 is still returning HTTP/1.1 403 Forbidden on /ubuntu/ | 18:21 |
| pior | utlemming, or on /ubuntu/dists/precise/universe/binary-amd64/Packages.bz2) | 18:22 |
| pior | s/)// | 18:22 |
| utlemming | pior: try "apt-get clean; apt-get -y update" | 18:22 |
| pior | utlemming, did it, apt-get update works 2/3 of the time | 18:23 |
| pior | # dig +short us-east-1.ec2.archive.ubuntu.com | 18:24 |
| pior | 10.252.111.96 | 18:24 |
| pior | 10.202.26.15 | 18:24 |
| pior | 10.250.142.223 | 18:24 |
| pior | utlemming, I'm working on the new stack for a couple of large deployments, so I'm not looking for a temporary solution | 18:26 |
| === wolferz_ is now known as wolferz | ||
| === wolferz is now known as Guest39644 | ||
| Soekris | Have some one running MAAS ? | 18:37 |
| ZenMaster | Hi guys, I was wondering if there was any support for the ATI 7000. | 19:14 |
| ZenMaster | I am trying to get some more performance out of my desktop on my Dell PowerEdge 2800. | 19:14 |
| Soekris | ZenMaster: have you a ubuntu desktop version running ? | 19:17 |
| ZenMaster | Soekris: Well "X" on Ubuntuer-Server. | 19:17 |
| Soekris | There is something called additional drivers | 19:18 |
| ZenMaster | I think i may have tried that let me look again;. | 19:18 |
| Soekris | This can install the closed software of the ATI driver | 19:19 |
| ZenMaster | Soekris: Ok I did install that. | 19:20 |
| ZenMaster | So what? do I go look for the fglr ATI driver and try to install this with it as a front end? | 19:20 |
| ZenMaster | Shows nothing when I open it. | 19:23 |
| Soekris | have you reboot you server ? | 19:24 |
| ZenMaster | I don't remember when the last time I did was. Has been months. I wonder if I installed that and never rebooted. | 19:26 |
| ZenMaster | Now taht would be funny. | 19:26 |
| ZenMaster | The machine says it needs to reboot everytime I login to a terminal. | 19:26 |
| ZenMaster | This is just a production server. | 19:26 |
| ZenMaster | Hard for me to do that. | 19:26 |
| Soekris | Oke | 19:26 |
| ZenMaster | Everyone is going to lunch here soon I can do it then. | 19:27 |
| Soekris | But the drivers from ati or nvidia have something else in the kernel | 19:27 |
| Soekris | so you must reboot | 19:27 |
| ZenMaster | My other question is though. I have never isntalled any ATI drivers or anything. | 19:27 |
| ZenMaster | This is a default install of X using proprietary driver. | 19:27 |
| ZenMaster | Should I try to install some kind of ATI driver? Is that why additional drivers shows nothing maybe? | 19:28 |
| Soekris | and if you have configured install security updates automaticaly you get also that message in the terminal | 19:28 |
| ZenMaster | Well I know that is part of the reason/ :) | 19:28 |
| ZenMaster | The Video in this is an ATI Radeon 7000-M. | 19:29 |
| Soekris | And if you have installed the additinal driver you must also reboot | 19:30 |
| akoumjian | Anyone else getting 403 forbidden from ec2 ubuntu archives? https://gist.github.com/2821127 | 20:44 |
| adam_g | zul: do you have any pending changes to glance-proposed or keystone-proposed that you haven't pushed yet? | 20:45 |
| pior | akoumjian, yes | 20:49 |
| pior | akoumjian, <utlemming> was supposed to look at this issue | 20:50 |
| akoumjian | pior: thanks. | 20:50 |
| pior | told me it was fixed : indead the second issue (one of the server in timeout) is fixed | 20:51 |
| pior | but waiting for it too | 20:51 |
| pior | 'utlemming> pior: we have new beta repos that you can use if you want. Run: sed -i "s,ec2.archive.ubuntu.com,ec2.archive.ubuntu.com.s3.amazonaws.com,g" /etc/apt/sources.list | 20:51 |
| akoumjian | pior: Again, thank you. I'm testing some automated deployments, so I will probably just wait until the issue is resolved. | 20:55 |
| === GhostFreeman_ is now known as GhostFreeman | ||
| === soren_ is now known as soren | ||
| jolaren | Any tips for nice programs for a server that's going to be hosted at work? I'm thinking PlaySMS for a SMS-gateway and Teamspeak/Socks5.. what more fun thing could you use a ION Asrock 330 for? | 21:24 |
| genii-around | Perhaps Quassel | 21:29 |
| jolaren | rather use irssi then but I don't need that | 21:31 |
| genii-around | Maybe a web-to-fax gateway | 21:35 |
| genii-around | ( or reverse, mgetty-sendfax in, emailed out ) | 21:36 |
| jolaren | that would be cool | 21:36 |
| jolaren | perhaps installing tor is also of intrest | 21:38 |
| === Adri2000 is now known as Guest35099 | ||
| xclusive585 | why is it, if I want to SSH in to my server and run a program, there is almost always problems with running something in the background (&) unless I run it "sudo"? | 22:13 |
| xclusive585 | I.E. some programs I use that have a command line server half, and a gui front end. I like the server half running in the background so I can just open GUI windows when necessary | 22:14 |
| xclusive585 | But if I do not use "sudo" and run whatever as root, it will exit when I close my SSH session. | 22:15 |
| fluvvell | I have a strange issue with a raid array, was running a swap partition on each of two drives in the array. Machine crashed with server panic, restarting rebuilds one of two arrays on drive and everything returns to normal. Smartctl reports both drives in array in full health. http://pastebin.com/K4G838kc | 22:35 |
| fluvvell | have since turned off one of the swap partitions, the one indicated in the error log - pasted log offending info from syslog. | 22:37 |
| fluvvell | the machine is running 10.04 server, 64bit and is up-to-date | 22:38 |
| xclusive585 | I've heard rumors about issues with system RAID volumes in 10.04. Sorry I cannot help here. | 22:41 |
| fluvvell | xclusive585: well its been running pretty flawlessly until a week ago | 22:45 |
| xclusive585 | So, does it rebuild everytime you reboot? or did it just do it once and then it was ok? | 22:56 |
| xclusive585 | and dont count on "smart" errors... SMART only catches an error in less than 50% of HDD fails. | 22:58 |
| xclusive585 | but considering it was just your swap partition that was acting up, I think the issue is more likely related to the kernel | 22:58 |
| fluvvell | xclusive585: ive only rebooted twice, and it rebuilt both times without my intervention | 22:59 |
| fluvvell | xclusive585: related to the kernel how? | 22:59 |
| fluvvell | twice in three weeks | 23:00 |
| xclusive585 | I just think it's weird because only the swap partition is rebuilding, and not the other(s)? | 23:02 |
| xclusive585 | that's why it seems like an OS issue and not the HDD. | 23:03 |
| fluvvell | xclusive585: I think a more accurate assessment of SMART would be that only in 50% of the time does SMART give you enough warning *before* a disk fails. Sorry, the swap partition is not a raid array - hope I did not mislead there. | 23:03 |
| xclusive585 | But I could be wrong. Honestly I'd test things by reinstalling Ubuntu, but I know that is not always an easy (or even possible) option | 23:03 |
| fluvvell | it is in fact md0 which is rebuilding, the root partition | 23:04 |
| xclusive585 | well, you only need one swap partition. So I'm confused how you had two swaps worked in together if they were not RAID | 23:04 |
| fluvvell | xclusive585: no, your right its not possible - its a production server. | 23:06 |
| fluvvell | xclusive585: http://tldp.org/HOWTO/Partition/setting_up_swap.html 9.3 documents more than one swap partition. Its not unusual | 23:06 |
| qman__ | looks to me like that drive is failing for some reason | 23:07 |
| qman__ | could be a loose cable or power issue just as easily as drive failure without SMART notice | 23:07 |
| xclusive585 | well, in my (short) experience, if you have say a two disk mirrored volume, then you set up two swaps (one on each) as a single mirrored swap vloume | 23:08 |
| qman__ | I wouldn't do that | 23:08 |
| xclusive585 | Thanks someone else. :-) I have no ideas here. | 23:08 |
| qman__ | no point | 23:08 |
| qman__ | your swap really shouldn't be containing critical data | 23:08 |
| xclusive585 | but how do you keep the two disks swaps equal for mirroriog if you dont? | 23:09 |
| qman__ | I guess it might work in preventing your server from crashing if a disk failed | 23:09 |
| qman__ | you don't | 23:09 |
| qman__ | it's a performance thing | 23:09 |
| qman__ | but unless you've got serious hardware, a disk failure often crashes the server anyway, or at least requires you to take it offline | 23:10 |
| xclusive585 | well, if you only create a swap on one disk, then you have either wasted space on the other, or two system partitions that aren't the same size, I guess is what im confused about | 23:10 |
| qman__ | you create a swap on both | 23:10 |
| qman__ | then it can use whichever is more convenient | 23:10 |
| fluvvell | qman__: one swap will be fine, its just I built the disks identically in layout so they both had a swap partition, I didn't need to swapon them both necessarily. | 23:10 |
| xclusive585 | so you just use two swaps then? | 23:10 |
| fluvvell | its just that errors from a swap partition are new to me | 23:10 |
| qman__ | fluvvell, what I'm saying is, the disk whose swap isn't working is probably failing | 23:10 |
| qman__ | make sure you've got a spare around | 23:11 |
| qman__ | I actually don't use swap, but that's neither here nor there | 23:11 |
| qman__ | RAM is cheap and fast | 23:11 |
| xclusive585 | my server never swapped more than a few k, lol and it has some 30 gigs of swap | 23:12 |
| fluvvell | qman__: yes, I had wondered that - is there a way to fsck or check the swap partn? or is badblocks enough? | 23:12 |
| qman__ | May 28 18:06:32 greenstone kernel: [609106.636488] end_request: I/O error, dev sda, sector 266662719 | 23:12 |
| xclusive585 | but in the future, I won't RAID the swaps, I'll just use two separate equal sized ones | 23:12 |
| qman__ | if it's consistently those sectors, it could be just a bad spot | 23:12 |
| fluvvell | qman__: yes I'm also familiar with the idea of using heaps of RAM instead, makes sense. | 23:13 |
| qman__ | but the odds are, since your array is rebuilding, that the disk is actually going bad or you have some other intermittent failure | 23:13 |
| fluvvell | qman__: ah, yes sector 266662719 - well spotted. | 23:14 |
| qman__ | xclusive585, it's fine to raid them if you want to avoid any downtime with a failure, since a failed disk with your swapped data on it would crash software, but like I said, without other serious hardware it'd probably crash anyway, a lot of disks bring down the controller when they fail and force you to edison boot | 23:14 |
| fluvvell | qman__: as it has crashed mine | 23:14 |
| fluvvell | qman__: edison boot???? | 23:15 |
| qman__ | fluvvell, kill the power, by either unplugging the cord or using the hard switch | 23:16 |
| qman__ | edison as in Thomas Edison and many power companies named after him | 23:16 |
| fluvvell | yes | 23:17 |
| fluvvell | qman__: Oh, power=edison, inventor - funny I've never heard the term | 23:17 |
| xclusive585 | I run soft raid, on my little home server. The idea is that no one drive failure causes a crash. So I guess I did ok. :-) I use a 2 disk mirrored volume, and a 4 disk RAID10 for data. Sure a failure would mean a reboot, but that's about it. | 23:17 |
| fluvvell | power companies not named that way in NZ | 23:17 |
| qman__ | xclusive585, then you did what you want, just be aware that you've still got coinflip odds it'll crash anyway | 23:18 |
| qman__ | in case of bad sectors it wouldn't, but in case of other types of failure it's not uncommon | 23:18 |
| xclusive585 | awww, now im scared. :) | 23:19 |
| qman__ | it's not permanent damage I'm talking about, just the drive controllers often lock up, so all your disks go offline | 23:20 |
| xclusive585 | as long as a replacement drive would allow it to boot back up and re-compile, I'm happy\ | 23:20 |
| qman__ | and if all your disks are on the same controller, chances are you have to power it off manually | 23:20 |
| qman__ | yeah | 23:20 |
| fluvvell | qman__: Problem: swap partition reports from fdisk that it starts at 266662935 | 23:20 |
| qman__ | fluvvell, right | 23:22 |
| qman__ | it's not necessarily in the swap partition, just that the failure caused the swap partition to become unreadable | 23:22 |
| fluvvell | Thats enough for me, I | 23:22 |
| fluvvell | I'm off to the hard drive shop :-) | 23:22 |
| qman__ | what you'd do to figure this out is attempt to read or write to those same sectors again to determine if it's spot related or not | 23:22 |
| qman__ | that's the best plan | 23:22 |
| fluvvell | now, with two identical drives, which is sda ? possibly the one in SATA0 I guess. | 23:23 |
| qman__ | you can get the serial number from smartctl | 23:23 |
| === Guest35099 is now known as Adri2000 | ||
| qman__ | the naming doesn't necessarily correlate to the physical connections in any way | 23:24 |
| xclusive585 | do what qman said, but typically yes, sda is sata0, sdb sata1 and so on | 23:24 |
| qman__ | so the serial number is how to be sure | 23:24 |
| qman__ | they're named in the order they're detected, which may or may not be the order they're hooked up, and may or may not be affected by other udev rules | 23:25 |
| qman__ | it's non-deterministic | 23:25 |
| fluvvell | qman__: well at least smartctl gives me the serial number :) | 23:26 |
| qman__ | that's why everything's moving to UUIDs | 23:26 |
| fluvvell | yes I use UUIDS | 23:26 |
| fluvvell | right thanks for the help, I'm off. | 23:31 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!