Kbase P21812: Database crashes when rfutil is proshut -F in a script
| Autor |
Progress Software Corporation - Progress |
| Acesso |
Público |
| Publicação |
29/06/2009 |
|
Status: Verified
SYMPTOM(s):
Database crashes when rfutil is proshut -F in a script
SYSTEM ERROR: mtlatch <latch-num>, holding <lock>. (3712)
SHUT 246: SYSTEM ERROR: mtlatch 2, holding 0x2. (3712)
SYSTEM ERROR: Releasing regular latch. latchId:<latch-num> (5028)
RFUTIL246: SYSTEM ERROR: Releasing regular latch. latchId: 1 (5028)
SHUT 246: SYSTEM ERROR: Releasing regular latch. latchId: 1 (5028)
User <num> died holding <num> shared memory locks. (2522)
User 246 died holding 1 shared memory locks. (2522)
SYSTEM ERROR: mtunlatch <latch-id>: not owner. (3714)
SYSTEM ERROR: mtunlatch 1: not owner. (3714)
Begin ABNORMAL shutdown code (2249)
crontab investigation shows two scripts that started at the same time as the database crash
core file created through _mprshut -F
dbx.out from _mprshut -F reads:
dsmFatalMsgnCallBack()
latXfree()
latpoplocks()
dbenvout()
dbUserDisconnect()
dsmThreadSafeEntry()
dsmUserDisconnect()
drexit()
drmsg.msgout()
msgnCB()
msgCB()
dsmFatalMsgnCallBack()
latlatch()
lockusr()
countUsers()
slactivity()
slmon()
Process running for _mprshut -F according to the ps -ef log
FACT(s) (Environment):
No errors in the system logs at the time of database crash
No delete of .lk file in cron jobs
Users connected in shared memory only
After-imaging switch every 10 minutes through cron jobs
CAUSE:
A cron job was running at the time which accesses promon activity to kill -2 a client. In this case the client was the rfutil client switching extents and this caused the database to crash.
FIX:
The rfutil connection is also classified as a client connection. The script that kills users off should exclude the rfutil PID's