Help! - 2 current logical logs
Posted in 2001
Topics: High Availability & Replication, Storage & Space Management, Error Codes & Troubleshooting, Logging & Checkpoints, Platform-Specific Issues
Hi all,
Please can someone save us?!
Our informix engine has got itself confused, and thinks it has 2 current
logical logs.
This is the bit from onstat -l :
15269500 20 U-B---- 380 33f7d5 20000 20000 100.00
1526951c 21 U-B---- 381 3445f5 20000 20000 100.00
15269538 22 U-B-C-L 382 349415 20000 20000 100.00
15269554 23 U-B---- 383 34e235 20000 20000 100.00
15269570 24 U---C-- 384 353055 20000 5240 26.20
1526958c 25 U-B---- 345 357e75 20000 20000 100.00
I've tried bouncing informix to see if that cleared it, but no, so now
it's stuck in fast recovery - this is the last bit from the message log
:
Informix Dynamic Server 2000 Version 9.21.UC3 -- Fast Recovery (CKPT
REQ) -- Up 22:49:02 -- 427296 Kbytes
Blocked:CKPT
13:58:42 Informix Dynamic Server 2000 Initialized -- Shared MemoryInitialized.
13:58:42 Physical Recovery Started.
13:58:42 Physical Recovery Complete: 5 Pages Restored.
13:58:42 Logical Recovery Started.
13:58:42 10 recovery worker threads will be started.
13:59:12 Waiting for Next Logical Log File to be Freed
Our LTAPEDEV is /dev/null. (so is TAPEDEV).
The platform is Linux RedHat 6.2, (kernel 2.2.16-3).
This server is a replicate of our live server - a CDR read-only target.
Before this happened, I'd had to shutdown CDR because we were getting
data corruption errors on the replicate and I had dropped & recreated
the replica db was about to reload it.
Unfortunately we have another db on this server that we want to keep,
and the only solution Tech support have given me so far is to
re-initialize the instance.
So what I'd really like to do is salvage my other db (I'm ashamed to
admit - there is no backup - agh!)
Does anyone have any ideas on how to do this ?
Would there be any clever hacks up Informix's sleeve to fiddle the logs
enough to recover, even temporarily ..?
Any ideas gratefully received.......
Here's the bit from the message log where it happened :
Fri Jan 19 00:04:11 2001
00:49:37 Fuzzy Checkpoint Completed: duration was 4 seconds, 3586buffers not flushed.
00:49:37 Checkpoint loguniq 382, logpos 0x4d31464
01:00:22 Logical Log 382 Complete.
01:00:22 Process exited with return code 142: /bin/sh /bin/sh -c/opt/informix/etc/log_full.sh 2 23 "Logical Log 382 Complete." "Logical
Log 382 Complete."
01:04:39 Fuzzy Checkpoint Completed: duration was 1 seconds, 4296buffers not flushed.
01:04:39 Checkpoint loguniq 383, logpos 0x937080
01:19:40 Fuzzy Checkpoint Completed: duration was 1 seconds, 5281buffers not flushed.
01:19:40 Checkpoint loguniq 383, logpos 0x194179c
01:30:53 Logical Log 383 Complete.
01:30:54 Process exited with return code 142: /bin/sh /bin/sh -c/opt/informix/etc/log_full.sh 2 23 "Logical Log 383 Complete." "Logical
Log 383 Complete."
01:34:42 Fuzzy Checkpoint Completed: duration was 3 seconds, 9987buffers not flushed.
01:34:42 Checkpoint loguniq 384, logpos 0x1477714
01:36:17 Assert Failed: No Exception Handler <details snipped>
01:36:25 Fatal error in ADM VP at mt.c:11380
01:36:25 Unexpected virtual processor termination, pid = 17955, exit =0x100
01:36:25 PANIC: Attempting to bring system down
17:47:55 Informix Dynamic Server 2000 Started.
Fri Jan 19 17:47:57 2001
17:47:57 Event alarms enabled. ALARMPROG =
'/opt/informix/etc/log_full.sh'
17:47:57 Booting Language <c> from module <>
17:47:57 Loading Module <CNULL>
17:47:57 Booting Language <builtin> from module <>
17:47:57 Loading Module <BUILTINNULL>
17:48:03 Informix Dynamic Server 2000 Version 9.21.UC3
17:48:10 Informix Dynamic Server 2000 Initialized -- Shared MemoryInitialized.
17:48:10 Physical Recovery Started.
17:48:10 Physical Recovery Complete: 1232 Pages Restored.
17:48:10 Logical Recovery Started.
17:48:10 10 recovery worker threads will be started.
17:48:12 Logical Recovery Complete. 19 Committed, 0 Rolled Back, 0 Open, 0 Bad Locks
17:48:14 Dataskip is now OFF for all dbspaces
17:48:15 Checkpoint Completed: duration was 1 seconds.
17:48:15 Checkpoint loguniq 382, logpos 0x215018
Hi, it's time to contact the Informix Technical Support ;-)) Or you simply patch the Flag of the Logical Log File with the unique id 382 back to U-B---L. Shouldn't be a problem if you would know, where you will find the appropriate flag in the current checkpoint page. I guess it's the fastest way to contact the support.... If you do not have a contract, Informix Support will help you when you really need them ! Sorry, Stefan