Re: [GENERAL] Update or Delete causes canceling of long running slave queries

2016-05-17 Thread Viswanath
Hi, What if we have a chain of slaves (master->slave->sec slave)? And if queries would run on each of them? or a config like master ->slave1 ->slave2 Is it ok to enable feedback on both slaves ? -- View this message in context: http://postgresql.nabble.com/Upd

Re: [GENERAL] Update or Delete causes canceling of long running slave queries

2016-05-14 Thread Jeff Janes
On Thu, May 12, 2016 at 9:22 PM, Viswanath wrote: > Hi, > Thank you for the explanation. > The slave queries are mostly reporting queries,which sometimes would take > 30+ running time due to complex joins and criteria. We haven't tried running > these queries on master before splitting, and it is

Re: [GENERAL] Update or Delete causes canceling of long running slave queries

2016-05-13 Thread Viswanath
Hi, Thank you for the explanation. The slave queries are mostly reporting queries,which sometimes would take 30+ running time due to complex joins and criteria. We haven't tried running these queries on master before splitting, and it is possible that there will be heavy updates and deletes on mast

Re: [GENERAL] Update or Delete causes canceling of long running slave queries

2016-05-12 Thread Jeff Janes
On Thu, May 12, 2016 at 11:14 AM, Viswanath wrote: > Hi Jeff, > Yes I am turning off autovacuum for experimental purpose only. I was > curious to know what is causing the queries to be killed when the autovacuum > is not running. I guess it had to be the pruning of HOT update chains like > you ha

Re: [GENERAL] Update or Delete causes canceling of long running slave queries

2016-05-12 Thread Viswanath
Hi Jeff, Yes I am turning off autovacuum for experimental purpose only. I was curious to know what is causing the queries to be killed when the autovacuum is not running. I guess it had to be the pruning of HOT update chains like you have mentioned. Also I have already tried changing the parameter

Re: [GENERAL] Update or Delete causes canceling of long running slave queries

2016-05-12 Thread Viswanath
Hi, I am using postgres 9.5 Yes,I restarted the server after changing the autovacuum to off. Also verified that autovacuum process is not running. standby db configurations: (mostly default conf only) hot_standby = on max_standby_archive_delay = 30s max_standby_streaming_delay = 30s wal_receiver_s

Re: [GENERAL] Update or Delete causes canceling of long running slave queries

2016-05-12 Thread Jeff Janes
On Thu, May 12, 2016 at 6:37 AM, Viswanath wrote: > Hi, > I have the following configurations. > > On master: > autovacuum=off > vacuum_defer_cleanup_age=0 > > On slave: > hot_standby_feedback = off > max_standby_streaming_delay = 30s > > Now when I try to run a long query on slave and do some up

Re: [GENERAL] Update or Delete causes canceling of long running slave queries

2016-05-12 Thread Sameer Kumar
On Thu, 12 May 2016 21:56 Viswanath, wrote: > Hi, > I have the following configurations. > > On master: > autovacuum=off > Have you changed it recently or it has always been off? In case you changed it recently, a restart would be needed for this setting to take effect. vacuum_defer_cleanup_age

[GENERAL] Update or Delete causes canceling of long running slave queries

2016-05-12 Thread Viswanath
Hi, I have the following configurations. On master: autovacuum=off vacuum_defer_cleanup_age=0 On slave: hot_standby_feedback = off max_standby_streaming_delay = 30s Now when I try to run a long query on slave and do some update or delete on the table on master I am getting the following error.