I've got three Linux systems (each with Fedora15-x86_64 running
PostgreSQL-9.0.4). I'm attempting to get a basic streaming
replication setup going with one master & two standby servers. At
this point, the replication portion appears to be working. I can run
an 'update' statement on the master, and view the result with a
'SELECT' on both standby servers.
I've mostly been working off the information presented here:
http://wiki.postgresql.org/wiki/Binary_Replication_Tutorial
http://wiki.postgresql.org/wiki/Streaming_Replication
plus the official Postgresql website documentation, with copious
googling for other sources when something didn't make sense.
0) I've successfully setup the WAL archiving on the master, and set
archive_timeout=61. However, what I'm seeing is that new files are
not getting generated every 61 seconds, but instead only when some
kind of SQL is invoked which writes to the database, or every 305
seconds (whichever comes first). I know this is the case because I
wrote a script which is being invoked via the archive_command option,
and that script is logging both the timestamp and filenames each time
it is invoked. How can I debug this?
1) Both of the wiki links above comment that the restore_command may
not be necessary if wal_keep_segments is large enough (mine is set to
128). I was going to setup the restore_command anyway, as I'm not yet
confident enough about streaming replication and failover with
postgresql to take chances, although the fact that i have two standby
servers makes this setup a bit more complex. However, can anyone
comment about whether its ever truly safe 100% of the time to run
without a restore_command ?
2) Can anyone who is currently running with more than one standby
comment how they're automating the failover process? I've read a few
comments about setting up something like heatbeat and/or pgpool (or
even writing custom scripts), but I'm not finding any well documented
solutions.
many thanks in advance!