[HPADM] vx_mapbad hpux 11.11
- From: Jeff Cleverley <jeff.cleverley@xxxxxxxxxxxxx>
- Date: Tue, 10 Jun 2008 17:06:49 -0600
We had a server generate an error with regards to vx_mapbad this morning. My best search attempts mostly came up with patches to fix the fsadm -d command. The other piece was the kernel error guide that said if it wasn't a I/O error to find out what caused the problem. Not very helpful :-)
The server is a rp5470 2x875mhz, 4 gig ram. It runs OnLine JFS 3.3 and hpux 11.11. The file systems are off of an EMC CX700 array direct connected via A6795A fibre cards. I checked the SP logs on the array and they don't report any type of errors anywhere near that time. This is the same configuration we've used for quite a while. We did some kernel changed around the Christmas break but we haven't seen any odd behavior from anything since then.
The behavior was the vx_mapbad (message 003) showed up in syslog and messages around 2:50 am. There were no other indicators of problems in any of the logs. We started getting calls from users because most systems could not mount this server or any of its file systems across nfs. You could manually mount the file system with no problem. When I logged into the server the load was less than .1, but the vxfsd was taking 101% of the cpu utilization. I was able to do some looking around at the file system. When I tried to unexport the file system, that prompt never came back. I did a find in the suspect file system which hung and the process would not die. This forced us to do a shutdown of the server. I had commented it from fstab and exports. When it came up I did a full fsck. No errors were reported and the file system mounted cleanly. The users of that data seem relatively happy other than the obvious and we have not heard of any corruption of data from the last couple of days of work.
Other than possibly installing PHKL_34805, I didn't see anything useful. The behavior of the file system and the system itself seems different than what is listed in the patch details. We did not do a TOC when the box was shut down and there is no core file. Naturally the system is no longer under software support.
Does anyone have any ideas on how we can try to isolate what really happened?
---> Please post QUESTIONS and SUMMARIES only!! <---
To subscribe/unsubscribe to this list, contact majordomo@xxxxxxxxxxxxx
Name: hpux-admin@xxxxxxxxxxxxx Owner: owner-hpux-admin@xxxxxxxxxxxxx
Archives: ftp.dutchworks.nl:/pub/digests/hpux-admin (FTP, browse only)
http://www.dutchworks.nl/htbin/hpsysadmin (Web, browse & search)
- Prev by Date: [HPADM] Backup Question
- Next by Date: [HPADM] Kernel : vmunix: file: table is full
- Previous by thread: [HPADM] Backup Question
- Next by thread: [HPADM] Kernel : vmunix: file: table is full