This is a discussion on Lock Without Owner within the Informix forums, part of the Database Server Software category; --> When I run onstat -k, I have found that I've got a lock without an owner. I've searched the ...
| |||||||
| FAQ | Members List | Calendar | Search | Today's Posts | Mark Forums Read |
| ||||
| When I run onstat -k, I have found that I've got a lock without an owner. I've searched the archives a bit and found 2 other threads (3+ years old) that mentioned something like this, but I never saw an explanation for it. I'm running on IDS 9.40.UC4 on SunOS 5.8. Here's a portion of the output showing the lock without the owner. sidb1p:informix[~]-[19]% onstat -k IBM Informix Dynamic Server Version 9.40.UC4 -- On-Line -- Up 171 days 14:23:24 -- 487424 Kbytes Locks address wtlist owner lklist type tblsnum rowid key#/bsiz a09713c 0 1221cb70 0 HDR+S 100002 205 0 <snip> 16cdd1b8 0 12230850 0 S 100002 205 0 16cde904 0 0 0 HDR+ 100002 206 0 16ce1ae4 0 12233f28 0 S 100002 205 0 <snip> 32 active, 16000 total, 2048 hash buckets, 3 lock table overflows sidb1p:informix[~]-[20]% I haven't noticed any issues with the database, but noticed this condition while investigating system status and thought it was a bit odd. Has anyone ever seen this? Is there an explanation for this lock? TIA, Gary |
| |||
| Gary Andrus schrieb: > When I run onstat -k, I have found that I've got a lock without an > owner. I've searched the archives a bit and found 2 other threads (3+ > years old) that mentioned something like this, but I never saw an > explanation for it. > Locks > address wtlist owner lklist type tblsnum rowid > key#/bsiz > 16cde904 0 0 0 HDR+ 100002 206 0 Can it be, that ER is running on your system ? I can reproduce this, when i define a replicate - it seems, that the log snooper cause the lock. Maybe Madison Pruet can explain this ? BTW: Check which database cause the lock: select database from sysmaster:systabnames where hex(partnum) = "0x00100002" -- it should be sysdatabases select * from sysmaster:sysdatabases where hex(rowid) = "0x00000206" -- syscdr ? Regards, try_and_err |
| |||
| Gary Andrus wrote: > When I run onstat -k, I have found that I've got a lock without an > owner. I've searched the archives a bit and found 2 other threads (3+ > years old) that mentioned something like this, but I never saw an > explanation for it. > > I'm running on IDS 9.40.UC4 on SunOS 5.8. Here's a portion of the > output showing the lock without the owner. > > sidb1p:informix[~]-[19]% onstat -k > > IBM Informix Dynamic Server Version 9.40.UC4 -- On-Line -- Up 171 > days 14:23:24 -- 487424 Kbytes > > Locks > address wtlist owner lklist type tblsnum rowid > key#/bsiz > a09713c 0 1221cb70 0 HDR+S 100002 205 0 > <snip> > 16cdd1b8 0 12230850 0 S 100002 205 0 > 16cde904 0 0 0 HDR+ 100002 206 0 > 16ce1ae4 0 12233f28 0 S 100002 205 0 > <snip> > > 32 active, 16000 total, 2048 hash buckets, 3 lock table overflows > sidb1p:informix[~]-[20]% > > > I haven't noticed any issues with the database, but noticed this > condition while investigating system status and thought it was a bit > odd. > > Has anyone ever seen this? Is there an explanation for this lock? > > TIA, > Gary > I've seen this after doing some nasty things in a session... namely onmode -z. I didn't investigate it... it was on a 10.00... After that I had to put the engine in single mode and this lock disappeared. Regards. |
| |||
| Thanks for the ideas. There's no ER running on the system, so that couldn't have been causing it. There were some onmode -z commands run, so that may have caused it, but I'm not sure. After the initial "sighting", I saw the same situation on a different server, on a different machine. I noticed that after some time, the locks on both machines went away. There was nothing in the logs to indicate any problems. |
| ||||
| The locks that you are indicating is the head of the shared locks on the database itself. The head of the database shared lock list is generally going to have an owner of zero. This is done to prevent a bogus deadly embrace on the locks associated with the database lock chain. FYI --- any lock on tblsnum 0x10002 is a lock on the database. "Gary Andrus" <andrusg@aetna.com> wrote in message news:1125416431.956476.113540@g44g2000cwa.googlegr oups.com... > When I run onstat -k, I have found that I've got a lock without an > owner. I've searched the archives a bit and found 2 other threads (3+ > years old) that mentioned something like this, but I never saw an > explanation for it. > > I'm running on IDS 9.40.UC4 on SunOS 5.8. Here's a portion of the > output showing the lock without the owner. > > sidb1p:informix[~]-[19]% onstat -k > > IBM Informix Dynamic Server Version 9.40.UC4 -- On-Line -- Up 171 > days 14:23:24 -- 487424 Kbytes > > Locks > address wtlist owner lklist type tblsnum rowid > key#/bsiz > a09713c 0 1221cb70 0 HDR+S 100002 205 0 > <snip> > 16cdd1b8 0 12230850 0 S 100002 205 0 > 16cde904 0 0 0 HDR+ 100002 206 0 > 16ce1ae4 0 12233f28 0 S 100002 205 0 > <snip> > > 32 active, 16000 total, 2048 hash buckets, 3 lock table overflows > sidb1p:informix[~]-[20]% > > > I haven't noticed any issues with the database, but noticed this > condition while investigating system status and thought it was a bit > odd. > > Has anyone ever seen this? Is there an explanation for this lock? > > TIA, > Gary > |