how to terminate global transactions?

[Frank Langelage]

On my Informix instance I have two global transactions holding some
locks. See onstat -G and onstat -x output below.
How can I get rid of them?
Also after a server reboot they are still there.
onmode -Z <address> or onmode -H <address> do not work (see below).

IBM Informix Dynamic Server Version 11.50.UC7IE on Solaris SPARC 10.

Global Transaction Identifiers
address flags isol timeout fID gtl bql data
56656bd8 -LX-G LC 0 131075 30 28
312D2D35336530656466663A613139383A34633139336231393A373461652D35336530656466663A613139383A34633139336231393A37346233
56656e28 -LX-G LC 0 131075 30 28
312D2D35336530656466663A613139383A34633139336231393A373463312D35336530656466663A613139383A34633139336231393A37346336
2 active, 128 total


Transactions

est.
address flags userthread locks begin_logpos current logpos
isol rb_time retrys coord
56655018 A---- 56623018 0 - -
COMMIT - 0
56655268 A---- 56623638 0 - -
COMMIT - 0
566554b8 A---- 56623c58 0 - -
COMMIT - 0
56655708 A---- 56624278 0 - -
COMMIT - 0
56655958 A---- 56624898 0 - -
COMMIT - 0
56655ba8 A---- 56624eb8 0 - -
COMMIT - 0
56655df8 A---- 566254d8 0 - -
COMMIT - 0
56656048 A---- 56625af8 0 - -
COMMIT - 0
56656298 A---- 56626118 0 - -
COMMIT - 0
566564e8 A---- 56626738 0 - -
COMMIT - 0
56656738 A---- 56626d58 0 - -
COMMIT - 0
56656988 A---- 56627378 0 - -
COMMIT - 0
56656bd8 -LX-G 0 2 logid 303 303:0x64b2622c
COMMIT 2:58 0
56656e28 -LX-G 0 2 logid 303 303:0x64b2922c
COMMIT 2:58 0
56657078 A---- 56627fb8 0 - -
COMMIT - 0
566572c8 A---- 56627998 0 - -
COMMIT - 0
56657e58 A---- 56629218 0 - -
COMMIT - 0
17 active, 128 total, 28 maximum concurrent

onmode -Z 1449487912

onmode: Cannot kill transaction 0x56656e28.
Only I-STAR subordinates that are PREPARE'd or HEURISTICally ABORT'd
may be heuristically completed.

onmode -H 1449487912

onmode -H may only be used to kill heuristically completed transactions.

[Davorin Kremenjas]

Post by Frank Langelage
On my Informix instance I have two global transactions holding some
locks. See onstat -G and onstat -x output below.
How can I get rid of them?
Also after a server reboot they are still there.
onmode -Z <address> or onmode -H <address> do not work (see below).

Hi Frank,

onmode will not work AFAIK. These global transactions were started by
a "distributed" application, (transaction manager) and Informix is
only a participant in XA transaction. Transaction manager (a wild
guess: Java app server?) needs to send either a commit or rollback now
to this participant.
Did your application crash? If it implements TP/XA transaction
management fully it should have means of finishing transactions in
progress after it comes back online.
If nothing works, call IBM Tech Support, they have a tool which will
clean this up for you.

HTH

Davorin

[Frank Langelage]

Post by Davorin Kremenjas

Post by Frank Langelage
On my Informix instance I have two global transactions holding some
locks. See onstat -G and onstat -x output below.
How can I get rid of them?
Also after a server reboot they are still there.
onmode -Z<address> or onmode -H<address> do not work (see below).

Hi Frank,
onmode will not work AFAIK. These global transactions were started by
a "distributed" application, (transaction manager) and Informix is
only a participant in XA transaction. Transaction manager (a wild
guess: Java app server?) needs to send either a commit or rollback now
to this participant.
Did your application crash? If it implements TP/XA transaction
management fully it should have means of finishing transactions in
progress after it comes back online.
If nothing works, call IBM Tech Support, they have a tool which will
clean this up for you.
HTH
Davorin

Thanks for your help.
Yes, it was a Java Application Server using distributed transactions and
this ran out of PermGen space so I had to kill -9 it.
Today I cannot figure out anymore why the transaction was not rolled back.
The instance is used for testing and developing and I do not have a
support contract for it (Innovator-C Edition).
Would dropping and recreating the database erase this open transaction?
Or do I have to initialize the whole instance?

Regards, Frank

[Davorin Kremenjas]

Post by Frank Langelage
Thanks for your help.
Yes, it was a Java Application Server using distributed transactions and
this ran out of PermGen space so I had to kill -9 it.
Today I cannot figure out anymore why the transaction was not rolled back.
The instance is used for testing and developing and I do not have a
support contract for it (Innovator-C Edition).
Would dropping and recreating the database erase this open transaction?
Or do I have to initialize the whole instance?

The transaction was not rolled back because Informix never got the
instruction to either roll it back or commit from your Java app
server. This is the correct behaviour accoring TP/XA specification.
Transaction managers should implement all the necessary logic to do
the rollback (or commit) after they come on-line after the crash. I
know JBoss does it.

The entry of this non-finished transaction is as always written to
your logical logs so dropping the database will not help (I'm not even
sure if you can drop it at all in the current state, with open
transaction on it, but you can always try).
Reinitialising the instance from scratch will certainly do the
trick...

HTH

Davorin

[Ulf]

Try onmode -l until you have forced a long transaction, this usually
help us. We have requested a way to get rid of these, but it seems
that the view of IBM that the TM should handle this i all cases.
Problem is that the TM fails sometimes, and then you are stuck.

Regards

Ulf

[Frank Langelage]

Post by Ulf
Try onmode -l until you have forced a long transaction, this usually
help us. We have requested a way to get rid of these, but it seems
that the view of IBM that the TM should handle this i all cases.
Problem is that the TM fails sometimes, and then you are stuck.
Regards
Ulf

Today, the transactions were rolled back by the server:

23:42:09 Aborting Long Transaction: tx: 0x56656bd8 no user info due to
XA or distributed (2-phase commit) transaction
23:42:09 Aborting Long Transaction: tx: 0x56656e28 no user info due to
XA or distributed (2-phase commit) transaction
23:42:13 Checkpoint Completed: duration was 4 seconds.
23:42:13 Thu Jul 1 - loguniq 306, logpos 0x3fff33b4, timestamp:
0xdedfaa19 Interval: 5616

23:42:13 Maximum server connections 15
23:42:13 Checkpoint Statistics - Avg. Txn Block Time 0.000, # Txns
blocked 1, Plog used 83432, Llog used 182271

23:42:13 Session completed abnormally. Rolling back tx id 24, flags
0x108463b
23:42:13 Session completed abnormally. Rolling back tx id 32, flags
0x108463b
23:42:13 Long Transaction 0x56656e28 Aborted. Rollback Duration: 0 Seconds
23:42:14 Long Transaction 0x56656bd8 Aborted. Rollback Duration: 0 Seconds

[j***@gmail.com]

Hi, the IBM tool to eliminate XA locks is called XAComplete. It requires that you attend a support engineer who knows and knows how to use.

[Pankrich k]

I think for the newer IDS version you can switch next logical log until over high watermark XA lock will be gone automatically. That's no need any tool.

For v7.3 if i rember well, it needs.


Pankrich Khanthasa-ard
By iPhone

Post by j***@gmail.com
Hi, the IBM tool to eliminate XA locks is called XAComplete. It requires that you attend a support engineer who knows and knows how to use.
_______________________________________________
Informix-list mailing list
http://www.iiug.org/mailman/listinfo/informix-list