[pgpool-general: 3811] Pgpool Watchdog and sharing information

[Jose Baez]

Hi!

In one of my scenarios DDBB servers need to shutdown from time to time
(servers are mounted on a vehicle which has to shutdown servers when
moving).

*Question is about failover* and "sharing status information" (with
master/master replication in PSQL).

*If PSQL node fails* (which is running pgpool with watchdog *in standby
mode* and a PSQL instance), the other PSQL node (with active Pgpool) will
detect that one PSQL node has failed and, at same time, will send the
Queries to the second PSQL node only.

As the first node is stopped, first pgpool's watchdog hasn't been  able to
receive the new status information.

After a while, every vehicle's server get shuts down.

When starting servers next time, if the first pgpool is the active one, it
won't know about the most updated status information: *it will see both
PSQL nodes as OK status*, won't it ? Which pgpool instance will run as
active?

Can i make both Pgpool to save/read the same "status_file" from a shared
network drive... would it work if there is only 1 pgpool instance at same
time? Does the file need specific permissions/owner?


Is there a robust Pgpool configuration which involves shutting down servers
often?  Would Streaming-Replication suits better?

Thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-general/attachments/20150616/75f33a28/attachment.html>