Enterprise Replication Problems on Linux Redhat 6.1

I am having major trouble getting ER to work for Informix 9.21 under Redhat Linux 6.1 I've opened a case with Informix, but am looking for some help from someone who's actually replicated under Linux before. I have setup 3 machines JEFF, MOE, LISA to replicate in UPDATE ANYWHERE mode a table called USER_OBJECTS. I'm just following instructions from the Guide to Informix Enterprise Replication, Replication Examples page 3-9 for the Update Anywhere example. My setup looks like this: SQL HOST FILE ON ALL MACHINES: ----------------------------- g_lisa group - - i=1 gigalisa onsoctcp lisa.gigaton.com informixjdbc g=g_lisa shmlisa onipcshm lisa.gigaton.com unique g=g_lisa jdbclisa onsoctcp lisa.gigaton.com informixjdbc g=g_lisa g_jeff group - - i=2 gigajeff onsoctcp 10.10.10.122 informixjdbc g=g_jeff g_moe group - - i=3 gigamoe onsoctcp moe.gigaton.com informixjdbc g=g_moe Replication Setup -------------- cdr define server --init g_moe cdr define server --connect=gigajeff --init --sync=g_moe g_jeff cdr define server --connect=gigalisa --init --sync=g_moe g_lisa cdr define replicate --conflict=ignore user_objects \\ "gigameta@g_moe:informix.user_objects" "select * from user_objects" \\ "gigameta@g_jeff:informix.user_objects" "select * from user_objects" \\ cdr change replicate --add user_objects \\ "gigameta@g_lisa:informix.user_objects" "select * from user_objects" cdr start replicate user_objects cdr list replicate user_objects REPLICATE SERVER STATE TABLE SELECT ---------------------------------------------------------------------------- user_objects g_moe ACTIVE user_objects select * from user_objectsuser_objects g_jeff ACTIVE user_objects select * from user_objectsuser_objects g_lisa ACTIVE user_objects select * from user_objects[10:45:05][informix@moe:1031] cdr list server SERVER ID STATE STATUS CONNECTION CHANGED ------------------------------------------------------------------ g_jeff 2 Active Connected Mar 27 19:56:45 2000 g_lisa 1 Active Connected Mar 27 22:43:19 2000 g_moe 3 Active Local [10:45:18][informix@moe:1032] SQL EXECUTED ------------- begin work; insert into objects values ('spiderman',1); insert into user_objects (username,password,oid,usedspace) values ('spiderman','spiderman','spiderman','1'); commit work; ERROR MESSAGE RECEIVED ON TWO MACHINE LOG FILES -------------------------------------------------- Tue Mar 28 10:38:22 2000 10:38:22 CDR CDRDS: transaction aborted (Error return from SQL for thecommit command) 10:38:28 Checkpoint Completed: duration was 0 seconds. 10:38:28 Checkpoint loguniq 53, logpos 0x25b018 10:43:27 Checkpoint Completed: duration was 0 seconds. 10:43:27 Checkpoint loguniq 53, logpos 0x25c018 I don't see any error messages for ATS/RIS either, only the above message. Doe anyone know why this is occurring? The transaction commits on the server I'm running the SQL from, but is not replicated to the other two servers. Any help would be appreciated. Thanks, Jeff Sent via Deja.com http://www.deja.com/ Before you buy.

Do you have a FK from user_objects to objects? Also, the reason you did not see any ATS/RIS entries is that you did not specify directories on your cdr define server statement. Redefine your servers and add the --ats,--ris to your replicate definition. Alan jeffhazelwood@my-deja.com wrote: > I am having major trouble getting ER to work for Informix 9.21 under > Redhat Linux 6.1 I've opened a case with Informix, but am looking for > some help from someone who's actually replicated under Linux before. > > I have setup 3 machines JEFF, MOE, LISA to replicate in UPDATE ANYWHERE > mode a table called USER_OBJECTS. I'm just following instructions from > the Guide to Informix Enterprise Replication, Replication > Examples page 3-9 for the Update Anywhere example. > > My setup looks like this: > > SQL HOST FILE ON ALL MACHINES: > ----------------------------- > g_lisa group - - i=1 > gigalisa onsoctcp lisa.gigaton.com informixjdbc > g=g_lisa > shmlisa onipcshm lisa.gigaton.com unique > g=g_lisa > jdbclisa onsoctcp lisa.gigaton.com informixjdbc > g=g_lisa> > g_jeff group - - i=2 > gigajeff onsoctcp 10.10.10.122 informixjdbc > g=g_jeff> > g_moe group - - i=3 > gigamoe onsoctcp moe.gigaton.com informixjdbc > g=g_moe> > Replication Setup > -------------- > > cdr define server --init g_moe > cdr define server --connect=gigajeff --init --sync=g_moe g_jeff > cdr define server --connect=gigalisa --init --sync=g_moe g_lisa > > cdr define replicate --conflict=ignore user_objects \\ > "gigameta@g_moe:informix.user_objects" "select * from user_objects" \\ > "gigameta@g_jeff:informix.user_objects" "select * from user_objects" \\ > > cdr change replicate --add user_objects \\ > "gigameta@g_lisa:informix.user_objects" "select * from user_objects" > > cdr start replicate user_objects > cdr list replicate user_objects > REPLICATE SERVER STATE TABLE > SELECT > ---------------------------------------------------------------------------- > user_objects g_moe ACTIVE user_objects > select * from user_objects> user_objects g_jeff ACTIVE user_objects > select * from user_objects> user_objects g_lisa ACTIVE user_objects > select * from user_objects> [10:45:05][informix@moe:1031] > > cdr list server > SERVER ID STATE STATUS CONNECTION CHANGED > ------------------------------------------------------------------ > g_jeff 2 Active Connected Mar 27 19:56:45 2000 > g_lisa 1 Active Connected Mar 27 22:43:19 2000 > g_moe 3 Active Local > [10:45:18][informix@moe:1032] > > SQL EXECUTED > ------------- > > begin work; > > insert into objects values ('spiderman',1);> > insert into user_objects (username,password,oid,usedspace) values > ('spiderman','s> piderman','spiderman','1'); > > commit work; > > ERROR MESSAGE RECEIVED ON TWO MACHINE LOG FILES > -------------------------------------------------- > Tue Mar 28 10:38:22 2000 > > 10:38:22 CDR CDRDS: transaction aborted (Error return from SQL for the> commit command) > 10:38:28 Checkpoint Completed: duration was 0 seconds. > 10:38:28 Checkpoint loguniq 53, logpos 0x25b018 > > 10:43:27 Checkpoint Completed: duration was 0 seconds. > 10:43:27 Checkpoint loguniq 53, logpos 0x25c018> > I don't see any error messages for ATS/RIS either, only the above > message. Doe anyone know why this is occurring? > > The transaction commits on the server I'm running the SQL from, but is > not replicated to the other two servers. > > Any help would be appreciated. > > Thanks, > > Jeff > > Sent via Deja.com http://www.deja.com/ > Before you buy.

Well, I worked with some of the Linux Platform Engineers during the testing of 9.21, so I guess that I qualify for having run 9.21 ER on the Linux platform. The error that you are getting occurs during the commit of the DataSync transaction on the target system. (The DataSync threads are the CDRD_### threads.) The replay of the transaction on the target system has been successful and is trying to commit and is running into some kind of problem. What it is we don't know because we aren't getting the ATS/RIS files created. If I'm correct you have opened a case (#946114) with tech support on this problem. I entered a note in the case earlier this evening. When you talk with Joe tomorrow, maybe you'll already be a bit further in a resolution. I'm going to relay in this news group much of what I put into the case. ER defers constraint checking until commit time. The reason for this is that some of your activies in the original trasaction might be shifted just a bit on the target instance. Therefor we do not want to run into referential constraint violations because we shifted the the insert of the parent row after the insert of the child. Since you are encountering the error during the commit phase, my first guess would be that there is some type of constraint violation which is causing the problem. But we need evidence of the actual error that you are getting so we have some idea as to resolve the problem. >From the case notes, I see that you modified the server to add the ATS/RIS directories. I'm assuming that you created this on all of the servers, that the directory exists, and that it can be written to by user informix. Did you also modify the replicates to include ATS/RIS spooling? To do this you would need to execute: cdr modify repl -A -R user_objects (There was a bug in this command in 7.31/9.2. --ris was mispelled. You had to enter --rts instead. I think this was corrected in 9.21.) After you have done this, you need to run your test again and see if the error is still occuring and if there is nothing in the ATS/RIS directories on the target system. If there is nothing there then here's what we will need for the case. 1) dbschema listing of the table definition on both the source and the target instances. Be sure to include the -ss option. 2) onstat -g cat 3) From syscdr select * from the following tables: a - repdef b - cdr_errors When you get this information, please pass it to the case owner. He will put it into the case. We can proceed from there. --------------------------------------------------------- Madison Pruet Informix Software Product Development - Enterprise Replication --------------------------------------------------------- jeffhazelwood@my-deja.com wrote: > I am having major trouble getting ER to work for Informix 9.21 under > Redhat Linux 6.1 I've opened a case with Informix, but am looking for > some help from someone who's actually replicated under Linux before. > > I have setup 3 machines JEFF, MOE, LISA to replicate in UPDATE ANYWHERE > mode a table called USER_OBJECTS. I'm just following instructions from > the Guide to Informix Enterprise Replication, Replication > Examples page 3-9 for the Update Anywhere example. > > My setup looks like this: > > SQL HOST FILE ON ALL MACHINES: > ----------------------------- > g_lisa group - - i=1 > gigalisa onsoctcp lisa.gigaton.com informixjdbc > g=g_lisa > shmlisa onipcshm lisa.gigaton.com unique > g=g_lisa > jdbclisa onsoctcp lisa.gigaton.com informixjdbc > g=g_lisa> > g_jeff group - - i=2 > gigajeff onsoctcp 10.10.10.122 informixjdbc > g=g_jeff> > g_moe group - - i=3 > gigamoe onsoctcp moe.gigaton.com informixjdbc > g=g_moe> > Replication Setup > -------------- > > cdr define server --init g_moe > cdr define server --connect=gigajeff --init --sync=g_moe g_jeff > cdr define server --connect=gigalisa --init --sync=g_moe g_lisa > > cdr define replicate --conflict=ignore user_objects \\ > "gigameta@g_moe:informix.user_objects" "select * from user_objects" \\ > "gigameta@g_jeff:informix.user_objects" "select * from user_objects" \\ > > cdr change replicate --add user_objects \\ > "gigameta@g_lisa:informix.user_objects" "select * from user_objects" > > cdr start replicate user_objects > cdr list replicate user_objects > REPLICATE SERVER STATE TABLE > SELECT > ---------------------------------------------------------------------------- > user_objects g_moe ACTIVE user_objects > select * from user_objects> user_objects g_jeff ACTIVE user_objects > select * from user_objects> user_objects g_lisa ACTIVE user_objects > select * from user_objects> [10:45:05][informix@moe:1031] > > cdr list server > SERVER ID STATE STATUS CONNECTION CHANGED > ------------------------------------------------------------------ > g_jeff 2 Active Connected Mar 27 19:56:45 2000 > g_lisa 1 Active Connected Mar 27 22:43:19 2000 > g_moe 3 Active Local > [10:45:18][informix@moe:1032] > > SQL EXECUTED > ------------- > > begin work; > > insert into objects values ('spiderman',1);> > insert into user_objects (username,password,oid,usedspace) values > ('spiderman','s> piderman','spiderman','1'); > > commit work; > > ERROR MESSAGE RECEIVED ON TWO MACHINE LOG FILES > -------------------------------------------------- > Tue Mar 28 10:38:22 2000 > > 10:38:22 CDR CDRDS: transaction aborted (Error return from SQL for the> commit command) > 10:38:28 Checkpoint Completed: duration was 0 seconds. > 10:38:28 Checkpoint loguniq 53, logpos 0x25b018 > > 10:43:27 Checkpoint Completed: duration was 0 seconds. > 10:43:27 Checkpoint loguniq 53, logpos 0x25c018> > I don't see any error messages for ATS/RIS either, only the above > message. Doe anyone know why this is occurring? > > The transaction commits on the server I'm running the SQL from, but is > not replicated to the other two servers. > > Any help would be appreciated. > > Thanks, > > Jeff > > Sent via Deja.com http://www.deja.com/ > Before you buy.