Good afternoon,
We run postgres 9.4 with hot standby streaming replication. We have been running abnormally high updates and the pg_xlogs have queued to about 80 percent of disk capacity. This often happens and the network is not fast enough to handle the transfer of logs to the replicated server.
What I have done in the past, and done again today is to change the archive_command from the script that scp's the logs to '/bin/true'. This stops the logship and allows the normal pg_xlog rotation to catch up.
This has not happened this time. Logs have been stored on pg_xlog since June 30 and the number of logs keep climbing. I am now manually moving some of the logs that were processed on June 30 off to another disk to alleviate the space. PLEASE HELP if you can.
It looks like archiving is still occurring despite '/bin/true' being set. Can I safely kill -term the archive thread?
Thank you for your prompt response.
--