pgbouncerin session pooling mode is absolutely transparent for applications, but it seems that it is not always true
osm2pgsql -> postgres-- ok
osm2pgsql -> pgbouncer -> postgres-- crazy errors
Hi all, I've started using PgBouncer with my team. We've run into an issue with hanging TCP connections when our database (AWS RDS instance) does a failover operation, meaning it doesn't send any packets over the wire back to PgBouncer to say that it's going away. At this point TCP connections get stuck until Linux kernel's
tcp_retries2 option kicks in (after about ~15mins).
More context here: https://github.com/pgbouncer/pgbouncer/issues/138#issuecomment-539842754
Would authors be open to me creating a PR with a solution to this?
2019-10-14 13:18:38.167 UTC  ERROR C-0x555c1e352ed0: unexpected SCRAM channel-binding attribute in client-final-message. SCRAM used to work just fine before I configured TLS. I also verified that TLS itself works correctly (certificates are valid and expected). Does anybody know what could cause the issue? Let me know if I can provide more information.
Channel binding is supported in PostgreSQL builds with SSL support. The SASL mechanism name for SCRAM with channel binding is SCRAM-SHA-256-PLUS. The channel binding type used by PostgreSQL is tls-server-end-point.(from https://www.postgresql.org/docs/11/sasl-authentication.html)
* = host=dbhostname port=5432. I used to have
default_pool_size = 1300, I've raised it to 2300 and have run reload. However, in one of the lines in
show databases;I still have
pool_size1300. How do I fix it without restart?
pool_sizedirective to the line in databases section and found it reported as a bug pgbouncer/pgbouncer#248 .
hi everyone, I have a problem using the query rewriting feature. I added in the in the file /etc/pgbouncer/pgbouncer.ini
rewrite_query_py_module_file = /etc/pgbouncer/rewrite_query.py
and in the file /etc/pgbouncer/rewrite_query.py this simple code:
def rewrite_query(username, query):
new_query = "select 1;"
I suppose that every query typed sould be changed to "select 1" but this doesn't occur. Any idea? thanks a lot
Hi Guys little help on graceful migration from one PostgreSQL cluster to another PostgreSQL cluster in Pgbouncer . After changing DB names in pgbouncer.ini I ran online restart og (-R) pgbouncer. After online restart and client restart some client connections (mostly read connections) are going to old RDS instances. Please take a look at issue
is there any other way of graceful migration with out down time? ( I mean to say other than RESET command)