Hello Postgres community,
I am running Postgres 9.0.3 on dual node RHEL 5.5 system. I have postgres setup to run on both nodes as active hot-standby. I am seeing a potential issue…or hopefully I have a configuration problem.
Every time I failover the DB I am seeing wal segement files (or checkpoint segment files) getting generated in the data/pg_xlog directory. It doesn’t seem like these files are being cleaned up and I am concerned that with enough failovers the file system will fill up and run out of space. Here are some of the details on the configuration that is being used:
wal_level = hot_standby
checkpoint_segments = 8
wal_keep_segments = 8
hot_standby = on
I don’t have archiving enabled. Config parameters look like the following:
# - Archiving -
#archive_mode = off # allows archiving to be done
# (change requires restart)
#archive_command = '' # command to use to archive a logfile segment
#archive_timeout = 0 # force a logfile segment switch after this
# number of seconds; 0 disables
This is the failover procedure I am using:
1. Touch the failover trigger file on the standby node
2. Stop the initially active DB
3. select pg_start_backup('clone',true) on the newly active DB
4. create backup tar up necessary files/directories under data directory, pull it over to node being recovered, and untar it in the data directory.
5. Start the postgres as the standby server.
6. select pg_stop_backup(); on the active DB
I am running this procedure repeatedly and new 16 meg files are being created in pg_xlog every time. Is this a know issue? Am I missing a step to ensure that file get recycled or deleted in pg_xlog? I did look in the archives but nothing jumped out at me. I appreciate any help that can be provided.
Thanks,
Matt Schoen