[pgpool-general: 6515] terminating connection due to idle-in-transaction timeout when using pgpool

[Jerameel Custodio]

Hello,

We are testing out pgpool, and overall it is working fine, but sometimes we
are running into idle-in-transaction timeouts. These same transactions are
able to complete without any issues when we are hitting the DB directly. If
we increase idle_in_transaction_session_timeout on the DB, it works, but
seems like a bandaid since we don't run into this without pgpool.

Our Postgres servers are on Amazon Aurora, and we are using pgpool for load
balancing only.

PostgreSQL version 9.6.9
pgpool-II version 3.7.8

Initially we had client_idle_limit set to 180, changed to 0, but did not
have any effect. We also tested with connection_cache on and off, with the
same result. The logs don't show much even with log level set to debug --
we can see the query logged, but nothing else from that PID afterwards, it
appears to silently exit once Postgres drops the connection due to the
timeout.

Just wanted to reach out to see if any high level insight or guidance can
be provided. If there is a better place for this, or if more details are
needed, let me know.

Thanks,
J
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.sraoss.jp/pipermail/pgpool-general/attachments/20190419/12f3f38a/attachment.html>