Reboot standby

[Thomas W. Ekberg]

We have 2 postgres servers, a primary and standby. One of our DB admins noticed a "health indicator" alert light on our standby which means "system degraded." She logged into the iLO and it lists the Cache Module Status as Failed. This needs to be repaired (we have an idea on what needs to be done) and most likely will require a reboot of our standby server.


We use streaming replication and from what I can see, that is initiated by the standby server. What are the shutdown and restart considerations for our standby server?



[1515078413450_PastedImage]

[Payal Singh]

As long as the streaming replication is in asynchronous mode, you can
restart the standby anytime without affecting the primary. You can check
whether replication is synchronous or asynchronous with "select sync_state
from pg_stat_replication;"

If it says 'synchronous', you will have to first change it to asynchronous
before you can reboot the standby, or else all commits on primary will be
pending waiting for the replica to replicate those commits first


Payal Singh,
Database Administrator,
OmniTI Computer Consulting Inc.
Phone: [phone] x 253

Post by Thomas W. Ekberg
We have 2 postgres servers, a primary and standby. One of our DB admins
noticed a "health indicator" alert light on our standby which means
"system degraded." She logged into the iLO and it lists the Cache Module
Status as Failed. This needs to be repaired (we have an idea on what
needs to be done) and most likely will require a reboot of our standby
server.
We use streaming replication and from what I can see, that is initiated by
the standby server. What are the shutdown and restart considerations for
our standby server?
[image: 1515078413450_PastedImage]

[Andreas Kretschmer]

Post by Thomas W. Ekberg
We use streaming replication and from what I can see, that is
initiated by the standby server. What are the shutdown and restart
considerations for our standby server?

your master has to keep enough wals that the standby can retrieve them
and catch up to the master. there are several solutions to achive that.
you can define wal_keep_segments or you can use replication slots. other
solutions: using archive_command to store the wals outside from the
master. you can also use barman as a wal-hub.


Regards, Andreas