topic Re: volinfo hangs. in Operating System - Tru64 Unix

volinfo hangs.

Christof Schoeman — Thu, 13 Oct 2005 09:08:11 GMT

Hi

volevac hangs on one of our production servers. Tracing the commands, we found that it is volinfo that causes the hangs. Running volinfo on its own also hangs.

We managed to reproduce the problem on a test system. It seems that volinfo hangs the moment you have more than 512 volumes in a group.

Has anyone seen this before, or know of a workaround?

Re: volinfo hangs.

Venkatesh BL — Thu, 13 Oct 2005 23:11:12 GMT

What is the OS/patch version?

If interested, you could see where exactly the command is hanging by using 'dbx'.

Re: volinfo hangs.

Christof Schoeman — Fri, 14 Oct 2005 01:21:46 GMT

The problem occurs on systems running V5.1B without patches, with PK4 and with PK5.

I am indeed interested in looking at the process using dbx.

Ran a truss on the volinfo and the vold processes. Here are the last few lines of each:

volinfo
-------
...
32331: write(3, " N E F V\0\0\0\0", 8) = 8
32331: sigaction(SIGPIPE, 0x000000011FFFBD90, 0x00000000) = 0
32331: read(3, "1CB004\0\0\0\0\0", 8) = 8
32331: read(3, " T S C V\0\0\0\0", 8) = 8
32331: read(3, "\0\0\0\0", 4) = 4
32331: read(3, "C7\0\0\0\0\0\0\0", 8) = 8
32331: read(3, "\0\0\0\0\0\0\0\0\0\0\0\0".., 307200) = 81912
32331: read(3, " e n\0\0\0\0\0\0\0\0\0\0".., 225288) = 81920
32331: read(3, "\0\0\0\0\0\0\0\0\0\0\0\0".., 143368) = 4112
32331: read(3, 0x000000014008B888, 139256) (sleeping...)

vold
----
...
73: munmap(0x0005E000, 8192) = 0
73: munmap(0x00060000, 8192) = 0
73: munmap(0x00062000, 8192) = 0
73: munmap(0x00064000, 8192) = 0
73: write(10, " N E F V\0\0\0\0", 8) = 8
73: gettimeofday64(0x000000011FFFBC98, 0x00000000) = 0
73: gettimeofday64(0x000000011FFFBC98, 0x00000000) = 0
73: sigprocmask(SIG_BLOCK, 0x00000000, 0x00000001400609C0) = 0
73: sigstack(0x00000000, 0x000000011FFFBC88) = 0

Re: volinfo hangs.

Venkatesh BL — Fri, 14 Oct 2005 03:05:43 GMT

On my machine, I see a "fork" call in volinfo and a "poll" call in vold...hmm...

Are you able to execute 'volprint'?

May be you could try 'voldctl disable' and then 'voldctl enable' to see if fixes the problems...but, read the manpage before proceeeding.

Re: volinfo hangs.

Stephen Wright_2 — Fri, 14 Oct 2005 04:11:08 GMT

Hi

voldctl disable/enable did not make a difference.

As for the fork you see in the truss output, I would assume your volinfo works and not not hang.

We created the 513 volumes with:
volassist make dsk0001 2M
on a test system and are able to reproduce the problem. If we delete a disk e.g
volume stop dsk0513
voledit -rf rm dsk0513
then volinfo works again.

Re: volinfo hangs.

Venkatesh BL — Fri, 14 Oct 2005 04:30:00 GMT

sounds like a bug to me. May be you should file a report with HP and get a resolution on this.

Re: volinfo hangs.

Stephen Wright_2 — Fri, 14 Oct 2005 08:30:04 GMT

Thank you

We have escalated to HP (some weeks now) and got a call ref no 1208733405.

Subsequent to escalating the call, we did our own tests and determined a volume limit of 512 per group if you want volinfo to work. Volprint works even if there are >900 volumes per group.