From: Brewer, Edward (BREWERE@OD.NIH.GOV)
Date: Mon Nov 04 2002 - 10:00:30 EST
More information to consider:
This is from the sqlplus user
Actually I did run a select statement from account synchronization job,
which <USER> and I were doing performance tuning. But before I left office
around 5:30pm last Friday, I already killed this job because I was
concerned it would impact deployment. But I didn't exit from sqlplus.
Following lists the statement I ran.
IRDB@IRDBPRD> SELECT <somestuff>
FROM <more stuff>
WHERE <even more stuff>
ERROR at line 9:
ORA-01013: user requested cancel of current operation
Elapsed: 00:21:42.11
<correction from previous submittal>
The next day we observerd that there was a process open for the above user.
There was a log file that was listed in lsof and the file exited on the
system, but it was not growing (and it was in the /usr/local/ filesystem...
we did a tail -f on it and found nothing writing to it, also it was small in
size...
Lee Brewer
Orginal Question:
Admins,
Something interesting happened this weekend. One of our file system,
/usr/local/ was approaching 100 percent at a steady pace. We assumed that
there must be an errant log file that was being written to. Using lsof and
fuser we couldn't locate any open files that were growing. Using du -axk |
sort -rn on the file system repeatedly gave no clue. We then shutdown our
databases and prepared for the worst. We noticed that there was a sqlplus
process that was charged to a user that was no longer logged in. We killed
the process and viola....the file system dropped to 3 percent..... Why were
we not able to see what this process had open. Has anyone ever seen this
before???
We are running Tru64 5.1 patch kit 5, TruCluster 5.1 on a GS160
Wildfire...
The file system is an advfs 4 domain with quota off...the process was
running on the server that serves the CFS filesystem
Lee Brewer
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:48:58 EDT