1

I have:

2017-04-19 09:33:26 UTC  :: WARNING:  page 43087 of relation base/11137505/11139144 is uninitialized
2017-04-19 09:33:26 UTC  :: CONTEXT:  xlog redo visible: rel 1663/11137505/11139144; blk 43087
2017-04-19 09:33:26 UTC  :: PANIC:  WAL contains references to invalid pages
2017-04-19 09:33:26 UTC  :: CONTEXT:  xlog redo visible: rel 1663/11137505/11139144; blk 43087

on 3 out of 7 slaves in replication on 9.3.9

The database 11137505 is not important for prod, so I decided to try using a broken db while building up a new slave and so I try to zero_damaged_pages = on to see what else is broken.

Yet it seems to ignore the setting appended to the end of the postgres.conf, producing the same log and terminating the instance.

This option is not meant for such usage - right? I can't make replicated slave to zero the corrupted page and go on replaying?

0

Your Answer

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

Browse other questions tagged or ask your own question.