9.6 write time
Hi, I am looking at a postgres 9.6 on rh7 I see that checkpoint_write_time is huge which looks quite strange as the average amount of data written is not that big. For example in 5:30 hours today, data from pg_stat_bgwriter view, comparing values at 11AM and 4h30PM : checkpoint_write_time 6986324 buffers_checkpoint 182447 so, to my understanding, it takes almost 2 hours to write 1.6 GB of data. Can someone either correct my understanding, or shed some light on what can cause this ?? thanks, Marc MILLAS Senior Architect +33607850334 www.mokadb.com
Re: 9.6 write time
Marc Millas writes: > I am looking at a postgres 9.6 on rh7 > I see that checkpoint_write_time is huge which looks quite strange as the > average amount of data written is not that big. checkpoint_write_time is not the amount of time spent doing I/O; it's the elapsed wall-clock time in the write phase. If the I/O is being throttled because of an un-aggressive checkpoint completion target, it could be a lot more than the actual I/O time. What have you got your checkpoint parameters set to? regards, tom lane
Re: 9.6 write time
Hi Tom, checkpoint completion target is 0.9 Marc MILLAS Senior Architect +33607850334 www.mokadb.com On Tue, Dec 28, 2021 at 6:46 PM Tom Lane wrote: > Marc Millas writes: > > I am looking at a postgres 9.6 on rh7 > > I see that checkpoint_write_time is huge which looks quite strange as the > > average amount of data written is not that big. > > checkpoint_write_time is not the amount of time spent doing I/O; > it's the elapsed wall-clock time in the write phase. If the I/O > is being throttled because of an un-aggressive checkpoint completion > target, it could be a lot more than the actual I/O time. What have > you got your checkpoint parameters set to? > > regards, tom lane >
Re: 9.6 write time
Marc Millas writes: > checkpoint completion target is 0.9 checkpoint_timeout is the more interesting number here. regards, tom lane
Re: 9.6 write time
Default ie. 5 minutes Le mar. 28 déc. 2021 à 19:34, Tom Lane a écrit : > Marc Millas writes: > > checkpoint completion target is 0.9 > > checkpoint_timeout is the more interesting number here. > > regards, tom lane >
