Cannot Rollforward from Checkpoint

Hi all, I have found frustrating the past six months trying to setup the HDR working again but the final recommendation from IBM was to migrate to the latest IDS version; we were doing fine with IDS 10.00.FC3 and made the mistake to upgrade to FC10. We are migrating from IDS 10.00.FC11 to 11.7.FC1. It was properly shutdown with onmode -ys; onmode -l; onmode -c ; onmode -yuk and have encountered the following issue: 05:33:51 IBM Informix Dynamic Server Started. 05:33:51 Requested shared memory segment size rounded from 1346560KB to 1376256KB 05:33:51 Shared memory segment will use large pages with intimate shared memory (ISM) if available 05:33:54 Segment locked: addr=10a000000, size=1409286144 Sun Oct 24 05:33:55 2010 05:33:55 Warning: ONCONFIG dump directory (DUMPDIR) '/IDS_logs/monterrey' has insecure permissions 05:33:55 Event alarms enabled. ALARMPROG = '/usr/informix/ids/etc/log_full.sh' 05:33:56 Booting Language <c> from module <> 05:33:56 Loading Module <CNULL> 05:33:56 Booting Language <builtin> from module <> 05:33:56 Loading Module <BUILTINNULL> 05:34:04 DR: DRAUTO is 0 (Off) 05:34:04 DR: ENCRYPT_HDR is 0 (HDR encryption Disabled) 05:34:04 Fast poll /dev/poll enabled. 05:34:04 IBM Informix Dynamic Server Version 11.70.FC1 Software Serial Number AAA#B000000 05:34:06 Conversion from version 10.00 Started 05:34:07 Checking partition header pages for needed free space 05:34:08 Checking for free space in partition header pages succeeded 05:34:08 RESTORE_POINT_DIR '/usr/informix/ids/tmp/10' contains old restore point files '/usr/informix/ids/tmp/10/reserved_pages' 05:34:08 Conversion control block initialization 05:34:08 Succeeded 05:34:08 Initialization of Conversion Session 05:34:08 Clearing out fuzzy-checkpoint-formatted physical log 05:34:15 Conversion waiting for physical recovery to complete 05:34:15 Initialization of Conversion Session Succeeded 05:34:15 Conversion Disabling Client Connections 05:34:15 IBM Informix Dynamic Server Initialized -- Shared Memory Initialized. 05:34:15 DR: Reservation of the last logical log for log backup turned on 05:34:15 Physical Recovery Started at Page (45:161). 05:34:16 Physical Recovery Complete: 0 Pages Examined, 0 Pages Restored. 05:34:16 Logical Recovery Started. 05:34:16 36 recovery worker threads will be started. 05:34:16 Open transaction detected when changing log versions. 05:34:17 Cannot Rollforward from Checkpoint. 05:34:18 oninit: Fatal error in shared memory initialization Now I don't have HA replication, no migration and more likely only another headache. Please guys, I need help !!!

Hi, I found the following in the documentation(http://publib.boulder.ibm.com/infocenter/idshelp/v115/index.jsp?to pic=/com.ibm.adref.doc/ids_adr_0776.htm): Open transaction detected when changing log versions. Cause The database server detected an open transaction while it was trying to convert the data from a previous version of the database server. Action Conversion is not allowed unless the last record in the log is a checkpoint. You must restore the previous version of the database server, force a checkpoint, and then retry conversion. Hope that will help, Regards

If that doed not work, IBM may be able to truncate the log to the previous checkpoint. Last resort is to bring up a clean new instances use archecker to extract your data from the archive and the logical logs taken after the archive. Art On Oct 24, 2010 10:12 AM, "MICHAEL KOESTER" <michael.koester@de.ibm.com> wrote: > Hi, > > I found the following in the > documentation( http://publib.boulder.ibm.com/infocenter/idshelp/v115/index.jsp?topic=/com.ibm.a dref.doc/ids_adr_0776.htm): > > Open transaction detected when changing log versions. > Cause > The database server detected an open transaction while it was trying to > convert the data from a previous version of the database server. > > Action > Conversion is not allowed unless the last record in the log is a checkpoint. > You must restore the previous version of the database server, force a > checkpoint, and then retry conversion. > > Hope that will help, > > Regards > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > --000e0cd28dba26b6ea0493605f72

I already did what IBM says, I think I'll have to make a call in order to truncate the logs. In that 10.00.FC11 instance I've been having a DR issue since Jan/09 and IBM recommendation was to migrate to latest version. I don't wanna follow the export/import upgrade :-(

Did you try going back to the previous version and backing up the logs ? I had a similar problem when going from 11.10 to 11.5. Just went back to 11.10 $INFORMIXDIR brought up db backed up logs brought down engine then went back to 11.5 $INFORMIXDIR brought up engine all ok. we use emterprise replication and I found I had to stop that as well before migrating.

I did, went back to 10.00.FC11, restore from ontape L0 twice, back'd up logical logs, and failed. The 2nd was to change all DBs to No-logging; onmode -sy ; onmode -c ; onmode -l ; onmode -c ; onmode -yuk and again IDS upgrade #Failed. I just saw pending transactions in systxptab but I cannot tell the begstamp date coz is not UNIX time as far as I know. If it is, is a long dead transaction (Jun 27,2010) and the logical logs gives a complete cycle in like 4 hours (1.2GB) in a regular busy OLTP day. Right now I put my staff to call tech support and see what's next to be done. I'll post it out the outcome.