Problem with distributed transactions

Hi, We are using an Informix 7.23.UC11 On-Line Database, on HP-UX 10.20 with MC-/Service Guard on two HP9000-K380, and TUXEDO v6.3. Following a crash on one server, due to a hardware problem, we use the second server for he high-availability of the application, we obtain one distributed transaction that we can't commit or rollback. The online.log is : ... 15:49:34 Physical Recovery Complete : 96 Pages Restored. 15:49:34 Logical Recovery Started. 15:49:38 Logical Recovery Complete. 97 Commited, 3 Rollback, 1 Open, 0 bad Locks. 15:49:38 Dataskip is now OFF for all dbspaces 15:49:38 On-Line Mode15:39:41 Checkpoint Completed: duration was 2 secondes. And "onstat -x" Informix-OnLine Version 7.23.UC11 -- On-Line -- Up 00:02:59 -- 16632 Kbytes Transactions adress flags userthread locks log begin isolation retrys coordinator c4903018 A---- c48df018 0 0 COMMIT 0 c490313c A---- c48df458 0 0 COMMIT 0 c4903260 A---- c48df898 0 0 COMMIT 0 c4903938 --X-G 0 26 3 COMMIT 0 c4903a5c A---- c48e0558 0 0 COMMIT 0 5 active, 128 total, 11 maximum concurrent. And the application can't be used because some locks exists on some data (as it is when a transaction is in duty), We try "onmode -Z c4903938", but this transactoins can't be destroyed. First, does anybody have an idea why the userthead is 0 ? Second, the address "c4903938" can't be find in the result of "onlog" !!! Actually THE ONLY WAY to restart the application is to ... RE-CREATE THE DATABASE !!! Have you any idea ? Thanks in advance.

In article <8fcm4h$rth$1@wanadoo.fr>, didier.sterbecq@wanadoo.fr writes >Hi, > >We are using an Informix 7.23.UC11 On-Line Database, >on HP-UX 10.20 with MC-/Service Guard on two HP9000-K380, >and TUXEDO v6.3. > >Following a crash on one server, due to a hardware problem, >we use the second server for he high-availability of the application, >we obtain one distributed transaction that we can't commit or rollback. > >The online.log is : > >... >15:49:34 Physical Recovery Complete : 96 Pages Restored. >15:49:34 Logical Recovery Started. >15:49:38 Logical Recovery Complete. > 97 Commited, 3 Rollback, 1 Open, 0 bad Locks. >15:49:38 Dataskip is now OFF for all dbspaces >15:49:38 On-Line Mode>15:39:41 Checkpoint Completed: duration was 2 secondes. > >And "onstat -x" > >Informix-OnLine Version 7.23.UC11 -- On-Line -- Up 00:02:59 -- 16632 Kbytes > >Transactions >adress flags userthread locks log begin isolation >retrys coordinator >c4903018 A---- c48df018 0 0 COMMIT 0 >c490313c A---- c48df458 0 0 COMMIT 0 >c4903260 A---- c48df898 0 0 COMMIT 0 >c4903938 --X-G 0 26 3 COMMIT 0 >c4903a5c A---- c48e0558 0 0 COMMIT 0 > 5 active, 128 total, 11 maximum concurrent. > >And the application can't be used because some locks exists on some data >(as it is when a transaction is in duty), > >We try "onmode -Z c4903938", but this transactoins can't be destroyed. > You are doing this onmode -Z on the machine which is the co- ordinator? Also check the $ONCONFIG file as this has no effect until TXTIMEOUT seconds have passed... Try this on both machines.. Then call Informix Techical Support. >First, does anybody have an idea why the userthead is 0 ? > >Second, the address "c4903938" can't be find in the result of "onlog" !!! > >Actually THE ONLY WAY to restart the application is to > ... RE-CREATE THE DATABASE !!! > >Have you any idea ? >Thanks in advance. > > -- David Williams