Disk Logging at boot

[Oviwan]

Hey folks

I have a case on oracle metalink and I should provide them the following information/logfile. For me it is easy to reproduce the case. But as they written, hpux doesn't have the logs they needed. Is there a way to enable a more precise loggin at booting?

From Metalink:
"From the investigation of the clusterware logs we have found that that issue most probably is a disk access issue, and to be certain from that we need to have the OS syslogs that cover the problem time found in the previous data collected part, but the issue is that the provided OS syslogs does not cover the problem date and time, so could you please provide us with the OS logs that covers the problem date and time to let us be able to continue the investigation."

Maybe it's a problem with OCR or VOTING disks.

My Env:
model:BL860c
OS:11.31
Packs: QPKAPPS B.11.31.0803.318a Applications Patches for HP-UX 11i v3, March 2008
QPKBASE B.11.31.0803.318a Base Quality Pack Bundle for HP-UX 11i v3, March 2008
Storage:EVA8100 with FC
I configured ocr and voting as alternate names:
e.g: mknod ora_ocr1_raw_1gb c 22 0x000008
and the permission are also good.

Thanks

Regards oviwan

[Jeeshan]

may be they are talking about event log.

[Pete Randall]

It sounds to me like the logs you gave them were fine, but they did not cover the date and time that they need to analyze. Double check the logs your self to see what time period they do cover.


Pete

[Rasheed Tamton]

Hi Oviwan,

>but the issue is that the provided OS syslogs does not cover the problem date and time, so could you please provide us with the OS logs that covers the problem date and time to let us be able to continue the investigation."


The oracle guys are asking you to give the old syslog files. Do you have log files in /var/adm/syslog which starts with OLD. That might be the ones they need. Just grep the OLD* for the date when the problem happened.

grep "Jun 18" OLD*
would give you the logs of 18th June.

Also check whether /var/stm/logs have anything.

Regards.

[Oviwan]

Thanks for reply.

I have already uploaded the OLDsyslog and syslog file.

this is the way I reproduce the issue.

I boot in single user mode and switch to runlevel 2 (init 2). In runlevel 3 clusterware start. then when I switch to runlevel 3 all the deamons etc... starts and after several seconds the server reboots and writes a dump. but there is no entry in syslog.

does any log exists about fibrechannel, disks, I/O's etc...? or which event logs do you mean, which achieve the requirements?

[Rasheed Tamton]

STM (Support Tool Manager) logs.

From STC doc.
-
If a device has a Failure, consult the device Tool Failure Log for the last active tool. This log gives likely causes for the failure and suggests possible actions. (If no Failure has occurred, no Failure Log will exist.)

To view a Failure Log, first select the device(s) affected, then choose the appropriate command:

xstm: Tools | Name_of_Tool_Detecting_Failure | Failure Log
mstm: Tools | Name_of_Tool_Detecting_Failure | Failure Log
cstm: infofailog, verfaillog, diagfaillog, exerfaillog, fwupdatefaillog, expfaillog, or utilfaillog.

--
rgds.

[Oviwan]

He wants to reconfigure syslogd, so that syslog logs everything.

this is the right way hm?
in /etc/syslog.conf
*.emerg;mail.none /var/adm/syslog/syslog.log

this is the highest level of logging?

[Oviwan]

I configured syslog.conf like this:
*.debug;mail.none /var/adm/syslog/syslog_debug.log

Thanks to all

regards