Bug #2351

DFBSD v3.1.0.579.g44ccf - Stuck during startup, random freezes

Added by tuxillo almost 2 years ago. Updated almost 2 years ago.

Status:In ProgressStart date:04/24/2012
Priority:NormalDue date:
Assignee:-% Done:

0%

Category:-
Target version:-

Description

Hi,

VM machine running DFly i386 with 1GB RAM / 1GB swap.

I'm experiencing random freezes that do not allow me to escape to DDB even. Following Sascha's recommendation I attached a virtual serial port to the VM and I was able to dump on a freeze during a shutdown connecting via serial port. After I rebooted the VM, I get this:

Mounting root from hammer:da0s1d
HAMMER(ROOT) recovery check seqno=04071621
HAMMER(ROOT) recovery range 3000000001c14c40-30000000053d4000
HAMMER(ROOT) recovery nexto 30000000053d4000 endseqno=0408dd14
HAMMER(ROOT) recovery undo 3000000001c14c40-30000000053d4000 (58454976 bytes)(RW)
HAMMER(ROOT) Found REDO_SYNC 3000000001c14c40
HAMMER(ROOT) recovery complete
HAMMER(ROOT) recovery redo 3000000001c14c40-30000000053d4000 (58454976 bytes)(RW)
HAMMER(ROOT) Embedded extended redo 3000000001c14c40, 0 extbytes
Warning: bio_page_alloc: memory exhausted during bufcache page allocation from swapper
Warning: bio_page_alloc: memory exhausted during bufcache page allocation from swapper
Warning: bio_page_alloc: memory exhausted during bufcache page allocation from swapper
Warning: bio_page_alloc: memory exhausted during bufcache page allocation from swapper
Warning: bio_page_alloc: memory exhausted during bufcache page allocation from swapper
Warning: bio_page_alloc: memory exhausted during bufcache page allocation from swapper
Warning: bio_page_alloc: memory exhausted during bufcache page allocation from swapper
[...]

Goes to the infinite, system is stuck printing those messages.

I'll try to gather more information, but in the mean time any advise is appreciated.

Cheers,
Antonio Huete

History

#1 Updated by tuxillo almost 2 years ago

  • Status changed from New to In Progress

I've added 1GB RAM more to the VM so that it avoids swapping and I've found it's doing some redo sync operation:

[...]
run record 3000000002488c00 objid 000000018d81b0cc offset 0000000000023e40 flags 00000001
run record 3000000002488e00 objid 000000018d81b0cc offset 000000000002b000 flags 00000001
run record 3000000002489000 objid 000000018d81b0cc offset 000000000002b1c0 flags 00000001
run record 3000000002489200 objid 000000018d81b0cc offset 000000000002b380 flags 00000001
run record 3000000002489400 objid 000000018d81b0cc offset 000000000002b540 flags 00000001
run record 3000000002489600 objid 000000018d81b0cc offset 000000000002b700 flags 00000001
run record 3000000002489800 objid 000000018d81b0cc offset 000000000002b8c0 flags 00000001
run record 3000000002489a00 objid 000000018d81b0cc offset 000000000002ba80 flags 00000001
run record 3000000002489c00 objid 000000018d81b0cc offset 000000000002bc40 flags 00000001
run record 3000000002489e00 objid 000000018d81b0cc offset 000000000002be00 flags 00000001
run record 300000000248a000 objid 000000018d81b0cc offset 000000000002bfc0 flags 00000001
run record 300000000248a080 objid 000000018d81b0cc offset 000000000002c000 flags 00000001
run record 300000000248a200 objid 000000018d81b0cc offset 000000000002c140 flags 00000001
[...]

After 20 minutes running it gets again the:

Warning: bio_page_alloc: memory exhausted during bufcache page allocation from swapper
Warning: bio_page_alloc: memory exhausted during bufcache page allocation from swapper
Warning: bio_page_alloc: memory exhausted during bufcache page allocation from swapper

#2 Updated by tuxillo almost 2 years ago

Assigning 4GB to the VM let the redo recovery process to finish (it run for 25 minutes).

[...]
unable to find objid 000000018d81b043:00040000
run record 30000000053d3800 objid 000000018d81b043 offset 00000000004dc800 flags 00000001
unable to find objid 000000018d81b043:00040000
run record 30000000053d3a00 objid 000000018d81b043 offset 00000000004dc9c0 flags 00000001
unable to find objid 000000018d81b043:00040000
run record 30000000053d3c00 objid 000000018d81b043 offset 00000000004dcb80 flags 00000001
unable to find objid 000000018d81b043:00040000
run record 30000000053d3e00 objid 000000018d81b043 offset 00000000004dcd40 flags 00000001
unable to find objid 000000018d81b043:00040000
HAMMER(ROOT) End redo recovery

I've also been able get the dump I could make while the system froze on shutdown.

It's on pkgbox32:/build/home/tuxillo/crash/*.1

This VM was migrated from an on old VMware Workstation v6.5-7 to 8. It had also a HAMMER v4 system which I upgraded to v6 recently. To discard any setup or VMware issues I'll reinstall latest i386 onto a new VM and check it out.

Also available in: Atom PDF