Conversion halted because logs are not backed up
Jacob Salomon got Conversion halted because logs are not backed up on 12.10 to 14.10 upgrades despite onmode -c and -l. Paul Watson asked about open transactions and ontape -a. Andreas Legner said the current log not being backed up should not cause it, so check onstat -l for other used logs. David Williams gave a shutdown sequence with log backup, stopping the scheduler and checking cluster sync.
Auto-generated by Claude from the posts below — may be imperfect; read the full thread.
Hi again.
My heels have not cooled since my last problem and I go another one. This keeps happening!
Before trying the upgrade we meticulously run
onmode -cand
onmode -land, as I watch the online log, I see a log backup start and finish. The we shut down the server and try the upgrade (12.10 => 14.10). In testing we often (it's inconsistent) get the message:
Conversion halted because logs are not backed up
.... As I compose my thoughts to write this, it occurs to me that even after a log switch and backup of the log file whence we just switched away, the new log is not backed up, although there is no activity to log at the time. I am adding to the upgrade document that the BA shall run onbar -c to back up the new (nearly empty) log.
Does this make sense to y'all who have gone through the pain of a release upgrade? Perhaps I just needed another set of eyes on our procedures.
Thanks for reading and commenting.
-- JS
------------------------------
+-----------------------------------------------------------+
| I am pleased to report that I had no problems today. |
| I had only issues, opportunities, challenges and valuable |
| learning experiences. |
+------------------------------------------ Jacob S --------+
------------------------------
Any open trans ? Expliciity run ontape -a?
------------------------------
Paul Watson
Oninit LLC
Cell +1 9133877529
------------------------------
The current log (C flag) not being backed up should NOT cause such message and failure, so there must be other used (U) logs missing the backup flag.
-> check onstat -l in such case (collect it before final shutdown preceding the upgrade attempt).
------------------------------
Andreas Legner
Informix Dev
HCL Software
------------------------------
Hi Jacob,
Convert the HDR to an RSS then it is not a fatal error!
Also stop the dbscheduler which can write to the log
https://www.ibm.com/docs/en/informix-servers/12.10.0?topic=saaf-scheduler-argument-stop-start-scheduler-sql-administration-api OR
touch $INFORMIXDIR/etc/sysadmin/stop before starting the instance
Shutdown sequence should be
Primary onmode -l && onmode -c && ontape -a && onmode -c
Answer "y" to backup the current log.
The last checkpoint flushes the backed up log status to the seondaries - IMPORTANT AND NOT IN THE MANUAL last time I looked!!!
onstat -g cluster to ensure all nodfes in sync
onstat -l on Primary AND Secondaries to ensure all logs markerd as backed up everywhere
onmode -ky Secondaries
onmode -ky Primary
Regards,
David.
------------------------------David Williams
Senior Database Platform Engineer
Flutter
London
------------------------------