Camelia, the Perl 6 bug

IRC log for #gluster-dev, 2013-10-18

| Channels | #gluster-dev index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:47 _ilbot joined #gluster-dev
01:47 Topic for #gluster-dev is now Gluster Development Channel - http://gluster.org | For general chat go to #gluster | Patches - http://review.gluster.org/ | Channel Logs - https://botbot.me/freenode/gluster-dev/ & http://irclog.perlgeek.de/gluster-dev/
02:03 bala joined #gluster-dev
03:19 kshlm joined #gluster-dev
03:31 bharata-rao joined #gluster-dev
03:42 kanagaraj joined #gluster-dev
03:52 itisravi joined #gluster-dev
04:11 spandit joined #gluster-dev
04:19 ndarshan joined #gluster-dev
04:42 lalatenduM joined #gluster-dev
04:44 ppai joined #gluster-dev
05:15 aravindavk joined #gluster-dev
05:21 mohankumar joined #gluster-dev
05:30 bala joined #gluster-dev
05:35 vshankar joined #gluster-dev
05:41 raghu joined #gluster-dev
05:44 ajha joined #gluster-dev
05:45 ababu joined #gluster-dev
05:45 kkeithley joined #gluster-dev
05:45 bfoster joined #gluster-dev
05:45 esalexa|gone joined #gluster-dev
05:46 hagarth joined #gluster-dev
05:47 portante joined #gluster-dev
06:16 lalatenduM joined #gluster-dev
06:20 lalatenduM joined #gluster-dev
06:20 spandit joined #gluster-dev
06:20 hagarth joined #gluster-dev
07:25 spandit joined #gluster-dev
07:32 hagarth joined #gluster-dev
07:44 hagarth left #gluster-dev
07:44 hagarth joined #gluster-dev
07:45 hagarth anybody facing problems logging in to gerrit?
08:22 itisravi joined #gluster-dev
08:50 spandit joined #gluster-dev
09:09 pzb joined #gluster-dev
09:12 kshlm joined #gluster-dev
09:17 vshankar joined #gluster-dev
09:38 kanagaraj_ joined #gluster-dev
09:46 itisravi joined #gluster-dev
10:20 jclift joined #gluster-dev
10:36 badone joined #gluster-dev
10:59 spandit joined #gluster-dev
11:10 kanagaraj joined #gluster-dev
11:42 rgustafs joined #gluster-dev
12:19 edward1 joined #gluster-dev
12:30 mohankumar joined #gluster-dev
12:31 Nev joined #gluster-dev
13:40 awheeler joined #gluster-dev
13:43 ndk joined #gluster-dev
14:27 hagarth joined #gluster-dev
14:53 bala joined #gluster-dev
14:58 jclift left #gluster-dev
15:01 vshankar joined #gluster-dev
15:02 bala1 joined #gluster-dev
15:18 mohankumar johnmark: ping
15:19 jclift joined #gluster-dev
15:19 jclift Well, my opinion of XFS just dropped from "good" to "sucks dogs balls".
15:20 kkeithley jclift: orly?
15:20 jclift It can't tell when someone is trying to mount an ext4 filesystem
15:21 jclift Eg, my laptop partitions got renumbered by my mucking around with partitions (unknown to me)
15:21 jclift Next boot, F19 attempts to remount one of my ext4 partitions onto /export/brick as an xfs filesystem
15:22 jclift Instead of barfing with a sensible error message like "This is not an XFS filesystem", it instead tells me my XFS filesystem is damaged and to run xfs_repair.
15:22 jclift Which I do.
15:22 jclift Long story short, I no longer have either a functioning ext4 filesystem either.
15:23 jclift Nor the data that used to be on it.
15:23 * jclift gives both F19 and XFS equally evil looks
15:24 johnmark mohankumar: pong
15:24 johnmark what's up?
15:24 johnmark jclift: lulz
15:25 kkeithley doctor, it hurts when I go like this. ;-)
15:25 johnmark kkeithley: then stop...
15:25 jclift Doctor shoots patient for being an idiot?
15:26 jclift Really, why can XFS not tell when someone is attempting to mount a non-xfs filesystem?  That seems kinda basic sanity checking.
15:26 jclift Meh, ignore that.
15:27 jclift It's basic, but there's probably a reason for it.  Something to do with "resources to fix it" and "caring".
15:27 kkeithley er, we have something like three or four people working on it.
15:27 kkeithley not gluster, Red Hat kernel fs team
15:28 jclift Yeah.  I guess they have other priorities.
15:28 jclift Checking when justin's paritions got renumbered and fstab is no longer pointing at the right filesystems is probably not high on their list.
15:28 jclift Hey, I'd like to be wrong there.  But I'm thinking I'm not. :D
15:29 kkeithley I just did a 'mount -t xfs /dev/sda6 /mnt' where /dev/sda6 is btrfs and it said "wrong fs type...."
15:30 jclift Try it with ext4 and see what happens?
15:30 * jclift took a screenshot of the error message
15:30 jclift It referenced specific lines of xfs .c code in it
15:30 jclift 1 min, I'll see if I can extract it from my phone.  Might actually turn out to be useful.
15:34 jclift kkeithley: http://justinclift.fedorapeopl​e.org/xfs_barfage/IMG_0462.JPG
15:35 jclift kkeithley: Worth filing a BZ?
15:35 johnmark mohankumar: you there?
15:37 kkeithley did init (or systemd) force the mount? I just tried to mount an ext4 with -t xfs and got the same "wrong fs type..." message. But yeah, I'd file a BZ
15:38 jclift kkeithley: No idea if it forced the mount.  The /etc/fstab option is set to "defaults".
15:38 foster that's syslog output
15:39 foster it's telling you why it couldn't mount the fs you told it was formatted as xfs ;)
15:39 kkeithley sorry, I misread your jpg. It's xfs_repair that didn't detect the ext4. IOW mount did the right thing and failed.
15:40 jclift Which thing said the "Corruption detected.  Unmount and run xfs_repair" ?
15:40 jclift So, that wasn't mount then?
15:41 jclift What part of the boot process was it, as I should probably file a BZ against that then?
15:41 kkeithley trying xfs_repair on my ext4 did the same thing yours did, i.e. didn't find the right magic number but proceeded anyway
15:41 kkeithley your mistake was following the directions to run xfs_repair
15:41 jclift k, so that's normal xfs_repair behaviour then.
15:41 kkeithley :-)
15:41 kkeithley not very friendly if it corrupted your ext4 anyway
15:41 foster I'm not sure, perhaps something is dumping the last few lines of dmesg on failure..?
15:42 jclift Hmmm.  My mistake was believing it when the filesystem utility told me it had detected corruption instead of saying "not an XFS filesystem".
15:42 foster that might not be a bug either, but perhaps it could include the mount command failure output to precede that output and provide some context
15:42 kkeithley all in all a sub-optimal experience
15:43 jclift Well, from "end users" perspective, something shouldn't be saying "Corruption detected." if it's not sure.  Obviously the normal command line version of "mount" detects this fine, why did [whatever the boot process uses] not do it right?
15:44 * jclift wonders if the boot process has some inspired use of --force options or similar hard coded in there
15:44 foster jclift: I suspect if you run "mount -t xfs /dev/sdxN /mnt" on anything not formatted xfs, it's going to print the same thing in the logs
15:44 jclift Ahhh
15:44 * jclift tries it
15:47 jclift http://fpaste.org/47798/21112231/
15:47 jclift foster: ^
15:48 foster read verification of the superblock is failing
15:48 foster (as expected)
15:48 jclift Seems like it dumps the last few lines of the system log output, minus the stack dump
15:49 foster right, so the mount output would probably be useful context to include in this scenario ;)
15:49 jclift Yeah
15:49 jclift foster: What should I file the BZ on?
15:49 foster i.e., enough of a hint that perhaps configuration is wrong
15:49 jclift eg which product/component do you reckon
15:49 jclift foster: Yeah
15:49 foster before you jump off the xfs_repair cliff ;)
15:50 foster hmmm, not really sure tbh
15:50 jclift Well, that wasn't the cliff that killed the data
15:50 jclift /export/brick is a temp gluster partition
15:50 foster systemd ?
15:51 jclift So, after xfs_repair started really taking a long time I just restarted the laptop + mkfs.xfs the thing again.  Problem solved.
15:51 jclift .... only to then find out that my partitions had been renumbered, and voila, I've nuked the wrong partition.
15:51 * jclift swears at misleading error message again
15:52 jclift foster: Thx.  I'll see if a BZ on systemd makes any difference
16:04 ndevos jclift: now update your /etc/fstab to mount by UUID, LABEL or use lvm and have decent /dev/vg_*/lv_* devicenames
16:08 jclift Yeah
16:11 jclift https://bugzilla.redhat.co​m/show_bug.cgi?id=1020981
16:11 glusterbot Bug 1020981: medium, unspecified, ---, systemd-maint, NEW , Misleading mount output from systemd can lead to tragic consequences.  Needs fixing.
16:12 wushudoin joined #gluster-dev
17:07 jclift Anyone know of an equivalent to fstyp for Linux?
17:14 kkeithley equivalent to what?
17:18 jclift fstyp
17:18 jclift It's a useful solaris command line utility.
17:18 jclift You run it like: fstyp /dev/sda6
17:18 jclift The it reports back the file of filesystem on the device.  eg xfs, ext4, etc
17:19 * jclift hasn't used Solaris in years, but it's one of the ones I wish had been ported to Linux
17:19 foster file -s
17:19 jclift Heh
17:19 jclift Exactly like that
17:20 jclift foster: http://oss.sgi.com/cgi-bin/gitweb.c​gi?p=xfs/xfs.git;a=commit;h=31625f2​8ad7be67701dc4cefcf52087addd88af4
17:21 jclift foster: Turns out the "Corruption detected" error is an XFS bug after all.  It was just spitting out that line or warning when it shouldn't have been
17:22 jclift Oh well, next update of xfs in F19 should have it fixed
17:22 foster jclift: I think that might refer to the situation where you do not specify an fs type
17:22 jclift Ahh.  Eric Sandeen just said it fixes the problem.
17:23 jclift Anyway, back to stuff I'm actually supposed to be working on... :)
17:23 foster I suppose it could be the case for bootup, I'm not really sure what's happening there
17:23 jclift np.  Thanks though, you've been super helpful with insight. :)
17:24 foster err maybe not, I tried it running a kernel that includes that change
17:24 foster Oct 18 13:23:54 localhost kernel: [ 6366.058532] XFS (vdb1): bad magic number
17:24 foster Oct 18 13:23:54 localhost kernel: [ 6366.060130] XFS (vdb1): SB validate failed with error 22.
17:31 jclift foster: Problem fixed, or problem not fixed?
17:34 foster well, if that message is printed in the probe case, that's definitely an xfs bug
17:35 foster it sounds like that's what was happening, so apparently the log messages above are what you'll see next time
17:35 foster as far as "bug fixed" goes, that's subjective I suppose
17:35 foster would the above have been intuitive enough for you?
17:36 foster it doesn't tell you to run repair at least :)
17:36 jclift Well, it's now indicating a more generic error message, so I would have prob figured out I'd stuffed my partition order.
17:36 jclift The "corruption detected" thing really led me to skip over that whole thing, and just fall into the trap of thinking I knew what was happening.
17:37 jclift I think the new wording is better.  Lets people know something is wrong, and they should be able to figure it out from there.
17:39 foster sounds reasonable to me
21:01 jclift left #gluster-dev
21:14 [o__o] joined #gluster-dev
22:05 awheeler_ joined #gluster-dev
22:46 badone joined #gluster-dev
23:43 [o__o] joined #gluster-dev

| Channels | #gluster-dev index | Today | | Search | Google Search | Plain-Text | summary