q4 procedure made system unstable.

[Jdamian]

Hi

Due to file lock problems, I found the HP doc KBRC00008493 (titled "sys adm: displaying the number of used file locks").

To get a list of all the locks, this doc suggests to run a Q4 procedure:

In my case, it produced errors and a file system became corrupted. Read the error messages in syslog.log (I removed unnecesary info to make shorter this list):

vmunix: msgcnt 1 vxfs: mesg 017: vx_trunc_typed - file system inode 4 marked bad
vmunix: ?vxfs msgcnt 1 offset 0x00000000 81b0 1 1f5 1f4
vmunix: ?vxfs msgcnt 1 offset 0x00000010 0 33fd400 404c82f6 8749c
vmunix: ?vxfs msgcnt 1 offset 0x00000020 404c82ff 407cb 404c82ff 455e8
vmunix: ?vxfs msgcnt 1 offset 0x00000030 30008 0 0 0
vmunix: ?vxfs msgcnt 1 offset 0x00000040 cffd 2b9c 0 1adb2b7f
vmunix: ?vxfs msgcnt 1 offset 0x00000050 1000000 0 1fac18 8
vmunix: ?vxfs msgcnt 1 offset 0x00000060 0 0 0 0
vmunix: ?vxfs msgcnt 1 offset 0x00000070 0 0 0 0
vmunix: ?vxfs msgcnt 1 offset 0x00000080 0 0 0 0
vmunix: ?vxfs msgcnt 1 offset 0x00000090 0 0 0 0
vmunix: ?vxfs msgcnt 1 offset 0x000000a0 0 0 0 0
vmunix: ?vxfs msgcnt 1 offset 0x000000b0 0 0


and /stand/vmunix is modified.
Then I tried to unmount that file system but umount command hung. I had to reboot the box.

What is happen ?

P.D: Q4 installed patches are:

# swlist -l product |grep -i q4
PHCO_28069 1.0 q4 patch version B.11.22l
PHCO_28601 1.0 q4 patch version B.11.23d

[Steve Steel]

Hi

Guess that it got coorupted in its buffers.

Reboot and mount cleared buffers .

It worked again.

There was probably a sleep due to a deadlock

If it happens again go to your response centre

Steve STeel

[Kent Ostby]

Hi --

I would suggest that something else is going on here.

Q4 does not modify things at all.

I suppose that its possible that if you run q4pxdb that it will modify the kernel.

I haven't seen in hundreds of runs where Q4 would modify the kernel or generate the messages you are seeing .

I'll look more into the document you site and the message listed above.

Best regards,

Kent M. Ostby

[Kent Ostby]

Okay .. the procedure did call for vmunix to be modified by q4pxdb so that explains the kernel modification.

The "msgcnt 1 offset" messages don't seem to be specific to Q4 although they could be caused by the command being run.

The only couple of documents that show these messages don't seem to indicate them as being problematic.

I believe you can ignore.

Best regards,

Kent M. Ostby