[12:19] <gpd> not as many ppl - which is also nice
[12:19] <daxxar> I see your point. :-)
[12:23] <gpd> you should repeat your question in here - they seem to know what they are talking about - although quiet
[12:24] <daxxar> Okay
[12:25] <daxxar> What could cause Samba to spontaneously reboot my system? I'm running an Ubuntu Server w/ Dapper RC.
[12:25] <daxxar> E.g. entering a subdirectory of my share just *reboots* the whole system. My syslog outputs this at the reboot
[12:26] <daxxar> May 28 00:12:40 datamania mountd[3869] : authenticated mount request from 192.168.0.14:937 for /storage/movies (/storage)
[12:26] <daxxar> May 28 00:15:45 datamania syslogd 1.4.1#17ubuntu7: restart.
[12:26] <daxxar> /storage is an LVM mount, it's also exported as NFS.
[12:28] <infinity> And it's also an automount?
[12:28] <infinity> samba's a red herring here, I think.  Samba would never cause your system to reboot.
[12:28] <daxxar> Yeah, it's an automount.
[12:28] <infinity> mountd could be tickling a kernel bug, though.
[12:28] <daxxar> Hm, okay
[12:31] <daxxar> Hmm. Yeah, I think you're right.
[12:31] <daxxar> I wasn't (previously) able to reproduce this in anything but the samba sessions.
[12:31] <daxxar> But I attempted a few IO operations via FTP, rebooted it there too now.
[12:31] <daxxar> Could it be my FS setup?
[12:31] <daxxar> It's setup for 'largefile' (1mb/inode)
[12:32] <daxxar> And data written right before the reboot is gone. (Directories created, files created)
[12:33] <daxxar> Just to have mentioned it, the memory passed a single full pass of memtest86. I didn't test it beyond that.
[12:33] <daxxar> Okay, it's just not that FS. Could it be my network-card?
[12:34] <infinity> I don't use automounts at all (haven't for years), so not sure how that could be affecting things.
[12:34] <infinity> Is it fine if you access non-mountd directories?
[12:34] <infinity> (Set up a temporary samba share on /tmp, for instance)
[12:34] <daxxar> No, accessing /home via FTP crashes it too.
[12:34] <daxxar> :o
[12:34] <daxxar> Uhm, automounts, is that the same as /etc/fstab-entries without 'noauto', or? :o
[12:34] <daxxar> (I assumed so when I answered)
[12:35] <infinity> Well, no.  See the "mountd" in your syslog?  That's not a samba thing.
[12:35] <infinity> I assumed you were using an automounter of some sort that was mounting filesystems on demand.
[12:36] <daxxar> Oh, nope.
[12:36] <daxxar> Or, shouldn't be.
[12:37] <daxxar> Hrrm. :|
[12:37] <daxxar> I guess this is HW-related.
[12:37] <infinity> Oh, wait.  mountd is the NFS automounter.
[12:37] <infinity> Silly me.
[12:37] <infinity> You have NFS exports on that machine as well?
[12:37] <daxxar> "dd if=/dev/zero of=test-big bs=1024K count=500" crashed it after ~300MB.
[12:37] <daxxar> Yes.
[12:38] <infinity> Yeah, okay.  The above syslog snippet was when connecting with NFS, not with samba.
[12:38] <infinity> And now I'm less confused.
[12:38] <daxxar> Ah, okay. :-)
[12:38] <infinity> Are you using nfs-user-server or nfs-kernel-server?
[12:38] <daxxar> kernel-server
[12:38] <infinity> Kay, that should be the more stable of the two.
[12:38] <infinity> Anyhow, you already confirmed that connecting some other way (like FTP) also crashed the box, yeah?
[12:38] <daxxar> Yep, and a dd via ssh.
[12:39] <infinity> Oh, you just said dd crashed.  I need to wake up a bit. :)
[12:39] <daxxar> So I Gess it's hw?: (
[12:39] <infinity> Fair chance it's hardware, then.
[12:39] <infinity> dd doesn't "just crash"  Ever.
[12:39] <infinity> It's far too simple to screw up.
[12:39] <daxxar> dd didn't *crash*, it made the box reboot. :-)
[12:39] <infinity> Well, same thing.
[12:40] <infinity> I would only expect dd to bugger the box if either A) you have hardware problems, or B) the kernel REALLY hates you.
[12:40] <infinity> Are you using an Ubuntu kernel, or hand-rolled?
[12:40] <daxxar> Ubuntu
[12:40] <infinity> Our kernels definitely shouldn't explode on filesystem access. :)
[12:40] <infinity> (This isn't reiserfs, is it?)
[12:42] <daxxar> Nope, ext3.
[12:42] <daxxar> Hm.
[12:42] <infinity> Kay, then I'd look at hardware for sure.
[12:42] <daxxar> Shutting down the nfs-kernel-server and the smb-server, and running a:
[12:42] <daxxar> daxxar@datamania:~$ i=1; while /bin/true; do echo "Pass: $i" && dd if=/dev/zero of=test-big bs=1024K count=50 && rm -f test-big && let i++; done
[12:42] <daxxar> Pass: 1
[12:42] <daxxar> I'll see if I manage to crash it now.
[12:44] <daxxar> Pass: 90. *me restarts it with count=100*
[12:46] <daxxar> Hmm. Now a single process hung, the rm -f test-big at pass #60. strace -p <pid> shows nothing on it.
[12:46] <daxxar> Can't kill it either.
[12:46] <frinkillo> maybe you should try running 'badblocks' to see if there's some damage on the HD
[12:46] <daxxar> Hrm. What piece of HW would you suspect is the most likely factor to be the cause?
[12:46] <daxxar> The HD was unwrapped ~6 hours ago. :-P
[12:47] <frinkillo> btw, hi ^^
[12:47] <gpd> SATA or SCSI ?
[12:47] <daxxar> IDE
[12:47] <daxxar> Hi. :-)
[12:47] <daxxar> Hmm. The console shows some huge amount of activity atm.
[12:47] <daxxar> Too fast to read.
[12:48] <frinkillo> I think you should try anyway to eliminate that possibility
[12:48] <daxxar> http://www.rafb.net/paste/results/z61UNQ99.html
[12:49] <daxxar>  x ~
[12:49] <frinkillo> sometimes brand new HDs come faulty from factory (very very rare, but happens)
[12:49] <daxxar> daxxar@datamania:~$ sudo init 6
[12:49] <daxxar> Segmentation fault
[12:49] <daxxar> ...
[12:49] <daxxar> :-P
[12:50] <frinkillo> wow
[12:50] <daxxar> But that's after the syslogmessages I pasted.
[12:50] <daxxar> It *did* say "a reboot is needed".
[12:51] <daxxar> "GRUB loading, please wait..." "Error 17"
[12:51] <Mysta> hey guys, I wanted to install vmware server beta and wanted to know if the server distro or the regular distro better suited me
[12:52] <frinkillo> daxxar: FS crashed?
[12:55] <frinkillo> Mysta: I think it doesn't matter as they are almost the same... but IIRC, vmware server comes with a gui, so it could be a better idea installing it on a regular (desktop) system
[12:56] <daxxar> frinkillo, yep. Seems so.
[12:56] <daxxar> Can't remount it via the rescue mode from the install CD.
[12:56] <daxxar> Any way to run badblocks without a full reinstall?
[12:56] <infinity> Then you either have a bad disk, or something in the CPU->Cache->RAM pipeline is horribly corrupting data before it gets to the disk.
[12:57] <infinity> (Or the driver for your controller sucks... Which controller is it?)
[12:57] <frinkillo> yeah, from the rescue CD, try something like 'badblocks -v -s /dev/xxx'
[12:57] <daxxar> The rescue-CD doesn't have badblocks, it seems. :|
[12:57] <frinkillo> ugh
[12:57] <daxxar> http://www.msi.com.tw/program/products/mainboard/mbd/pro_mbd_detail.php?UID=15 - That mobo
[12:57] <daxxar> 
[12:57] <daxxar>  Intel ICH2 Chipset Chipset
[12:57] <daxxar> Onboard ATA controller.
[12:58] <infinity> Oh, that's very well supported.
[12:58] <daxxar> Mkay. You saw the error i pasted on rafb.net?
[12:58] <infinity> So, you're looking at a bad disk, or corruption on the way to the disk (bad CPU/cache, bad RAM, or bad motherboard)
[12:58] <daxxar> How many passes of memtest86 would be needed to rule out RAM-problems? I ran a full pass, no errors.
[12:59] <infinity> Yeah, that didn't tell me much.  Bad paging request could be any of the above as well.
[12:59] <daxxar> Okay.
[01:00] <infinity> Has that CPU ever been overclocked, by chance?
[01:00] <Mysta> anyone/
[01:00] <infinity> Mysta: If you want to use the GUI, go for a desktop install.
[01:01] <infinity> Mysta: If you want the server kernel on a desktop install, just "apt-get install linux-server"
[01:02] <daxxar> infinity, underclocked. :o
[01:02] <daxxar> (AFAIK)
[01:02] <daxxar> It should be a 1.2GHz P3, but it's clocked at 1.0GHz
[01:02] <infinity> daxxar: And it's never been overclocked in the past?
[01:03] <Mysta> thx guys
[01:03] <daxxar> Not AFAIK. But this CPU was inherited from someone else. ;-)
[01:03] <infinity> daxxar: Only curious because this smells like "burnt cache", which is a common side-effect of overclocking.
[01:03] <daxxar> (For some reason, I can't get the multiplier to 9 which it should be, only to 8)
[01:03] <infinity> daxxar: If that someone else was underclocking it to compensate for overclocking it in the past, that might say something. :)
[01:03] <daxxar> Mkay. Doesn't memtest86 test cache too?
[01:03] <infinity> Not reliably, but it tries.
[01:04] <daxxar> Okay. Should I use the Ubuntu Dapper Desktop to test the disk, or get the Seagate tool?
[01:04] <infinity> You can disable cache in the BIOS and try to reproduce the crashes, but it'll run REALLY SLOW without a cache.
[01:04] <infinity> Seagate's tool is more likely to find real problems on the physical disk.
[01:05] <infinity> badblocks will only find what's exposed to the ATAPI/IDE layer, which should be "nothing" on a modern disk, since the disk's firmware is supposed to be swapping our bad blocks on the fly.
[01:05] <daxxar> Hm, okay.
[01:05] <infinity> (Not that this works as well in practice as it's meant to in theory)
[01:06] <frinkillo> hmm interesting
[01:07] <infinity> s/swapping our/swapping out/
[01:07] <infinity> I make more sense with fewer typos.
[01:09] <daxxar> It made sense before the regreplace too. ;)
[01:11] <daxxar> SeaTools includes a RAM-test. :p
[01:13] <Mysta> infinity: i was searching aptitude for that server kernel by searching for linux-server, and came up short, is that the correct name?
[01:13] <daxxar> Ugh. 1 hour to run a full scan of the disk. :|
[01:14] <infinity> Mysta: Yeah.  "linux-server" is probably in restricted, though.  If you only have "main" enabled, "linux-image-server" should do.
[01:14] <Mysta> k, thx
[01:14] <infinity> Mysta: If you're on breezy and not dapper, ignore everything I've said, there are no server-tuned kernels on breezy.
[01:15] <daxxar> Thanks a lot for the help frinkillo, infinity, gpd, :-)
[01:15] <Mysta> infinity: ok, thats why. lol. I just realized i ssh'd to my breezy instead of dapper
[01:15] <daxxar> I'll let this disktest from Seagate run, see if it finds anything.
[01:15] <daxxar> If not, I'll run memtest86 for a few hours.
[01:16] <frinkillo> :)
[01:16] <daxxar> If not, I'll tear my hair out. :|
[01:16] <daxxar> If nothing there, *
[01:16] <Mysta> is there any documentation on what is different in this kernel compared to a regular one?
[01:17] <Mysta> nvmd, i found something
[01:17] <frinkillo> https://wiki.ubuntu.com/ServerFaq
[01:18] <Mysta> that helps too thx
[02:34] <eimajenthat> the description says "development discussions," is there a general ubuntu-server channel?
[02:35] <daxxar> Okay.
[02:35] <daxxar> "Full test" completed, no errors.
[02:35] <daxxar> So the problem is not controller / disk. *sighs*
[02:36] <eimajenthat> the description says "development discussions," is there a general ubuntu-server channel?
[02:36] <daxxar> I bet people actually could see what you said the first time.
[02:46] <daxxar> infinity, do you think the error could be caused by a defective PSU?
[02:47] <daxxar> infinity, it doesn't seem to be HD, controller or memory. Could be CPU or MB I guess.
[02:47] <daxxar> (defective or not powerful enough. It should be 300W.)
[04:33] <Zambezi> Is it possible to make Ubuntu-server (Breezy) with to modification, more secure? And how? I would like as high security as possible. Please PM or highlight me name so I won't miss an answer.
[12:14] <phace> well how can i get involved with the ubuntu server development ?
[02:29] <daxxar> Hm, how strange.
[02:29] <daxxar> I ran a full pass on memtest86 before installation, no errors. Then, I ran two passes last night before I went to bed, no errors.
[02:30] <daxxar> Today, I started a pass, and it found 3 errors before it had run for 5 minutes.
[02:31] <daxxar> If it said it found it at address ~122MB, is it safe to assume there are errors on the memorychip in the first slot, since it's 256MB?
[02:31] <daxxar> (The 'channel' field of the errors is blank)
[02:48] <infinity> It's not really save to assume anything, unless you know for sure how the memory controller is interleaving the RAM.
[02:48] <infinity> s/save/safe/
[02:49] <infinity> Your best bet it to do repetitive memtest runs on the machine witch each stick installed individually.
[02:49] <infinity> Pain in the ass, but it's the only reliable way to find a single bad stick.
[02:50] <daxxar> So I need to run how many passes on each stick?
[02:50] <daxxar> Since this was found 5 minutes into the 4th actual pass.
[02:50] <daxxar> = ~5 hours.
[02:50] <daxxar> Eh, 4.
[02:52] <infinity> RAM errors come and go, depending on how well it decides to hold a charge from one moment to the next.
[03:22] <kermitX_> new default phpmyadmin theme sux.. no width contstraints. inputs on far left & submit on far right...