cancel
Showing results for 
Search instead for 
Did you mean: 

The server crashed

hy_5
Occasional Advisor

The server crashed

One day,all the members of the alpha 4000 cluster crashed.In the messsage file,there are lots of messages like below:
Jan 11 02:25:19 qd1 vmunix:
Jan 11 02:25:19 qd1 vmunix: simple_lock: time limit exceeded
Jan 11 02:25:19 qd1 vmunix:
Jan 11 02:25:19 qd1 vmunix: pc of caller: 0xfffffc000059ea58
Jan 11 02:25:19 qd1 vmunix: lock address: 0xfffffc007fe4f668
Jan 11 02:25:19 qd1 vmunix: lock info addr: 0xfffffc00009771d0
Jan 11 02:25:19 qd1 vmunix: lock class name: cam_softc
Jan 11 02:25:19 qd1 vmunix: current lock state: 0x8000011d005a5301 (cpu=0,
pc=0xfffffc00005a5300,busy)
Jan 11 02:25:19 qd1 vmunix:
Jan 11 02:25:19 qd1 vmunix: panic (cpu 0): simple_lock: time limit exceeded
Jan 11 02:25:19 qd1 vmunix: syncing disks... device string for dump = SCSI 1 200
0 0 0 0 0 0.
Jan 11 02:25:19 qd1 vmunix: DUMP.prom: dev SCSI 1 2000 0 0 0 0 0, block 2048000
Jan 11 02:25:19 qd1 vmunix: device string for dump = SCSI 1 2000 0 0 0 0 0.
Jan 11 02:25:19 qd1 vmunix: DUMP.prom: dev SCSI 1 2000 0 0 0 0 0, block 2048000
What is the matter?thank you.

And there are some other error messages:
Jan 10 01:07:06 qd1 vmunix: malloc failed: bucket size = 256, #of failures = 1,
ra 0xfffffc000042e124
Jan 10 04:10:05 qd1 vmunix: ADVFS EXCEPTION
Jan 10 04:10:05 qd1 vmunix: Module = ms_logger.c, Line = 1886
Jan 10 04:10:05 qd1 vmunix: release_dirty_pg: log half full
Jan 10 04:10:05 qd1 vmunix: panic (cpu 1): release_dirty_pg: log half full
Jan 10 04:10:05 qd1 vmunix: syncing disks... device string for dump = SCSI 1 200
0 0 0 0 0 0.
Jan 10 04:10:05 qd1 vmunix: DUMP.prom: dev SCSI 1 2000 0 0 0 0 0, block 2048000
Jan 10 04:10:06 qd1 vmunix: device string for dump = SCSI 1 2000 0 0 0 0 0.
Jan 10 04:10:06 qd1 vmunix: DUMP.prom: dev SCSI 1 2000 0 0 0 0 0, block 2048000
6 REPLIES
Ravi_8
Honored Contributor

Re: The server crashed

Hi,

Looks like CPU 0 is in problem, Log a call with HP
never give up
hy_5
Occasional Advisor

Re: The server crashed

Thank you.But I don't think so.Because there is 'vmunix: panic (cpu 1): release_dirty_pg: log half full'.
Is it a bug?
Han Pilmeyer
Esteemed Contributor

Re: The server crashed

Which version and patch kit are you running?

"log half full" dumps were quite common in V4. Can't remember seeing them with V5, at least not often.

The 2nd issue (on the 11th) has a lock timeout on cam_softc. If one had to take a guess at which hardware would be at fault, it would be some storage related device.
Michael Schulte zur Sur
Honored Contributor

Re: The server crashed

Hy,

I remember from installing patches that there was a reference to a simple lock panic. What os and patch kit are you running?

greetings,

Michael
hy_5
Occasional Advisor

Re: The server crashed

OS is digital 4.0D,and no patch kit.
Han Pilmeyer
Esteemed Contributor

Re: The server crashed

At the very least I would install the latest V4.0D patch kit. Even better would be to upgrade to V4.0F with its latest patch kit. This might reduce the chance of getting a "log half full" panic, although I don't believe that this was ever fully resolved in the V4 stream.