1

We have a master server and replication server running Postgres 9.3.6

Streaming was going slow due to a secondary failover router. It was running behind but still working.

I got a communication error and now I am getting the 

ERROR:  requested WAL segment 00000001000003150000001E has already been removed

However looking at the master server that file (and many before it) is still in the WAL directory.

I restarted the replication server and I do get the message 

redo starts at 315/1A9CB5A0
[4018 - 2015-09-28 20:57:32 GMT]LOG:  consistent recovery state reached at 315/1ED5E910
[4018 - 2015-09-28 20:57:32 GMT]LOG:  unexpected pageaddr 313/C0D60000 in log segment 00000001000003150000001E, offset 14024704

Is there any way to restart the streaming or will I have to pull down and do another basebackup?

Improve this question
5
  • Are there messages after the "unexpected pageaddr" line?
    – Kassandry
    Commented Sep 28, 2015 at 22:11
  • "the WAL directory". Do you mean pg_xlog? Or an archive directory?
    – Craig Ringer
    Commented Sep 29, 2015 at 0:42
  • Messages after the pageaddr line are the "ERROR: requested WAL segment 00000001000003150000001E has already been removed". WAL directory yes the pg_xlog. I think there was more going on with the network than they admitted at first which created the issue. We have passed our WAL archive limit for restarting so I will need to take another base backup to start things again. However...if something such as this happens in the future is there a way to move the WAL files to the replication server locally to have them replay/locate there or how would this situation be handled? Thanks.
    – Toolman21
    Commented Sep 29, 2015 at 15:42
  • Are you sure your are connecting to the master server you think you are connecting to, the one that has the file in its pg_xlog? No amount of past network problems (assuming they are no longer occurring) should prevent the master from handing over the file if it still has it.
    – jjanes
    Commented Sep 29, 2015 at 17:30
  • I should have clarified I suppose. The network issue I think had been and still was blocking at the time communication so I was dead in the water as it turns out (different from my original post). It thought it was communicating, but it was not...at least not completely. I could have however moved the WAL files manually if there was an option to have the replication server read WAL locally. Just to keep it going until the network issue was corrected.
    – Toolman21
    Commented Sep 29, 2015 at 18:33

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.