Обсуждение: Cascade publication with pg_stat_replication empty
I have a server which replicates using Publication/Subscription. On subscriber server I have some ETL processes running on it and from this second server I´ve created a new publication to send this database to a third server. While this second server was copying data with COPY to third, everything was fine but when COPY finished and it would start streaming data, both replication just stopped. View pg_stat_replication is empty on first server and subscriber server is not connected to it. Exactly same occurs on second, its pg_stat_replication is empty and subscriber, the third server, is not connected to it. But WAL is being generated on first server, which seems that it is waiting the second to connect to it. So, pg_publication are filled correctly on publishers but pg_stat_replication is empty on both First and second servers are version 11.11 and third one is 13.0 -- Sent from: https://www.postgresql-archive.org/PostgreSQL-general-f1843780.html
On Wed, 2021-04-07 at 04:16 -0700, PegoraroF10 wrote: > I have a server which replicates using Publication/Subscription. On > subscriber server I have some ETL processes running on it and from this > second server I´ve created a new publication to send this database to a > third server. While this second server was copying data with COPY to third, > everything was fine but when COPY finished and it would start streaming > data, both replication just stopped. Look into the PostgreSQL log on both machines. Yours, Laurenz Albe
On publisher it shows me "terminating walsender process due to replication timeout" But the problem is that this message is occurring 3 or 4 times a day only. It tries to send data, gets a timeout then shows the message. Then it waits a long long long time to try again, I don´t know why. wal_sender_timeout is configured to 1 hour Yesterday I had 4 messages of timeout on server log but no data were replicated. So, it tried just these 4 times ? -- Sent from: https://www.postgresql-archive.org/PostgreSQL-general-f1843780.html
So, there is a way to call replication, to client restart connection and try again ? Because aparently it stays disconnected for hours to try again, gets that timeout message and remains disconnected for hours again -- Sent from: https://www.postgresql-archive.org/PostgreSQL-general-f1843780.html