Archive log not flushing after full database backups

[mladams]

ANR3265W .. database backups (continuously) are being triggered due to the archive log space over the threshold. This has been going on for 14 hours since the first full database started last night at 11pm and finished at 1am this morning. I read an article where a full database backup crossing over midnight into the next day could cause this issue at 7.1.1 but we are server level 8.1.5. Does a TSM halt/restart flush the archive log along with the active log?

 

[marclant]

You may have a long running transaction pinning the logs.

The oldest active log gets archived once all the transactions in it are committed. If you have multiple active log files, the oldest transaction is the one holding the oldest log. If you halt the server, there's no more inflight transactions, active log returns to 0.

Archive log is only pruned after a successful DB backup.

 

[mladams]

Thank you for your response. So if I'm understanding this correctly, the only way to fix the archive log is to somehow remove the transaction that is pinning the logs and then run a full database backup. So a server halt will return the actlog to zero and then upon a server restart, I should run a full database backup to prune the archive log?

 

[marclant]

o if I'm understanding this correctly, the only way to fix the archive log is to somehow remove the transaction that is pinning the logs

No, that deals with the active log.

then run a full database backup.

To clear the archive log only.

So a server halt will return the actlog to zero and then upon a server restart, I should run a full database backup to prune the archive log?

In theory yes, in practice you should not have to do that. You might have one, many or all of the following problems:
- slow client with long transactions
- slow DB disk causing long transactions, also long DB backups.
- slow active log disk, slowing down DB updates, causing long transactions
- not enough active log space if it fills up
- not enough archive log space if it fills up between DB backups.

 

[Stephan]

Do you have some unneeded archive logs in your archive folder? Maybe there are archive logs dating back from a few days that were not cleaned up somehow and are taking quite a bit of space? I ran into an issue once, where i found logs that dated back from a few days and where i'd had FULL backups run successfully in between. I simply deleted those logs.

Or you may simply need to increase your logging space. Additional nodes, longer processes etc may cause higher logging.

 

[RecoveryOne]

Do you have some unneeded archive logs in your archive folder? Maybe there are archive logs dating back from a few days that were not cleaned up somehow and are taking quite a bit of space?

I've had that happen as well with various versions of the server from time to time as well.

 

[mladams]

Thanks for all your help. I never received a response from IBM. Interesting as an NDMP backup failure cause the database to go into an evaluation mode. After the evaluation mode, the /tsmarc log started to clear out. A subsequent server halt/restart followed by a full database backup further cleared out the log.