Hello. I've had several instances where postgres on my physical replica
under version 9.6.6 is crashing with messages like the following in the
logs:
2018-04-18 05:43:26 UTC dbname 5acf5e4a.6918 dbuser DETAIL: The postmaster
has commanded this server process to roll back the current transaction and
exit, because another server process exited abnormally and possibly
corrupted shared memory.
2018-04-18 05:43:26 UTC dbname 5acf5e4a.6918 dbuser HINT: In a moment you
should be able to reconnect to the database and repeat your command.
2018-04-18 05:43:26 UTC dbname 5acf5e39.68e5 dbuser WARNING: terminating
connection because of crash of another server process
2018-04-18 05:43:26 UTC dbname 5acf5e39.68e5 dbuser DETAIL: The postmaster
has commanded this server process to roll back the current transaction and
exit, because another server process exited abnormally and possibly
corrupted shared memory.
2018-04-18 05:43:26 UTC dbname 5acf5e39.68e5 dbuser HINT: In a moment you
should be able to reconnect to the database and repeat your command.
2018-04-18 05:43:27 UTC 5acf5e12.6819 LOG: database system is shut down
When this happens, what I've found is that I can go into the pg_xlog
directory on the replica, remove all the log files and the postgres will
restart and things seem to come back up normally.
So, the question is what's going on here... is the log maybe getting corrupt
in transmission somehow? Should I be concerned about the viability of my
replica after having restarted in the described fashion?
Thanks,
Greig Wise
--
Sent from: http://www.postgresql-archive.org/PostgreSQL-general-f1843780.html
