My newly upgraded opensolaris 2008.11 laptop crashed last weekend.
(The OS was installed from the os 2008.11 live-cd and then upgraded
using the package manager to snv_111.)
I was trying to copy a large virtual pc image from my wife''s imac to
the
laptop. On a whim I had decided to create a separate zvol in the
root pool to contain the image, figuring I could create a vbox
with linux or windows to run vmware''s conversion tool (not even
knowing which os it runs on).
I made several mistakes with this.
1. I did not get the size of the image right and
created a zvol of 24G instead of 25.
2. I enabled the time-slider
3. I did not disable autosnapshots on the zvol
(enabled by default, as inherited from the root pool)
4. My laptop has a lowly 32bit processor, which I''ve learned
has some trouble with atomic 64bit operations.
Looking back, I had set up a little disaster waiting to happen,
but not this one - the laptop spontaneously panic''ed.
The reboot was quick, but shortly after restarting the rsync
I started getting warnings that the rpool was running out of space.
In no time, time-slider had removed all auto-snapshots and kept
complaining about an 80% fill rate of the rpool.
I''m not sure about the root cause of the actual crash.
It looks like the time-slider-cleanup that was just started
from cron was in some way responsible.
* Here are the last two lines of cron before the crash:
> CMD: /usr/lib/time-slider-cleanup -y
> root 9383 c Sat Apr 18 22:35:00 2009
* This is the log of the crash itself:
Apr 18 22:35:17 glorantha unix: [ID 836849 kern.notice]
Apr 18 22:35:17 glorantha ^Mpanic[cpu1]/thread=db542aa0:
Apr 18 22:35:17 glorantha genunix: [ID 403854 kern.notice] assertion
failed: vp->v_count > 0, file: ../../common/fs/vnode.c, line: 817
Apr 18 22:35:17 glorantha unix: [ID 100000 kern.notice]
Apr 18 22:35:17 glorantha genunix: [ID 353471 kern.notice] d6d44e34
genunix:assfail+5a (fea9f89c, fea9f930,)
Apr 18 22:35:17 glorantha genunix: [ID 353471 kern.notice] d6d44e54
genunix:vn_rele+7e (d84c1ac0, 0, 0, fea)
Apr 18 22:35:17 glorantha genunix: [ID 353471 kern.notice] d6d44f34
genunix:vn_renameat+384 (0, 80474c0, 0, 80b3)
Apr 18 22:35:17 glorantha genunix: [ID 353471 kern.notice] d6d44f64
genunix:vn_rename+18 (80474c0, 80b309c, 0)
Apr 18 22:35:17 glorantha genunix: [ID 353471 kern.notice] d6d44f84
genunix:rename+13 (80474c0, 80b309c, 8)
Apr 18 22:35:17 glorantha unix: [ID 100000 kern.notice]
Apr 18 22:35:17 glorantha genunix: [ID 672855 kern.notice] syncing file
systems...
Apr 18 22:35:17 glorantha genunix: [ID 904073 kern.notice] done
Apr 18 22:35:18 glorantha genunix: [ID 111219 kern.notice] dumping to
/dev/zvol/dsk/rpool/dump, offset 65536, content: kernel
Apr 18 22:35:29 glorantha genunix: [ID 409368 kern.notice] ^M100% done:
81098 pages dumped, compression ratio 2.35,
Apr 18 22:35:29 glorantha genunix: [ID 851671 kern.notice] dump
succeeded
Apr 18 22:35:55 glorantha genunix: [ID 540533 kern.notice] ^MSunOS
Release 5.11 Version snv_111 32-bit
Apr 18 22:35:55 glorantha genunix: [ID 943908 kern.notice] Copyright
1983-2009 Sun Microsystems, Inc. All rights reserved.
Apr 18 22:35:55 glorantha Use is subject to license terms.
Apr 18 22:35:55 glorantha unix: [ID 126719 kern.info] features:
21567fff<cpuid,mwait,cmp,sse3,nx,sse2,sse,sep,pat,cx8,pae,mca,mmx,cmov,de,pge,mtrr,msr,tsc,lgpg>
Apr 18 22:35:55 glorantha unix: [ID 168242 kern.info] mem = 1046588K
(0x3fe0f000)
* And the cronlog resumed 40s later:
! *** cron started *** pid = 276 Sat Apr 18 22:36:32
2009> CMD: /lib/svc/method/zfs-auto-snapshot
svc:/system/filesystem/zfs/auto-snapshot:frequent
If anything, this crash was *smooth*. If I hadn''t been sitting
in front of the laptop, I wouldn''t have noticed that it happened.
(Except that my login session (and rsync) disappeared, of course)
If anyone''s interested, I saved the dump...
But I wonder if this could be a case of bug 6634371 (not so atomic
64 bit operations on 32bit cpu)? Time to get a new laptop...
Cheers,
Henk Langeveld
