too frequent refresh about diskquota HOT 7 CLOSED

Do you mean diskquota should not refresh until receiving a signal, and the gap time between two signals should be larger than diskquota_naptime? But the signal is also required to be sent at regular intervals to keep the data in diskquota latest, isn't it?

from diskquota.

Do you mean diskquota should not refresh until receiving a signal, and the gap time between two signals should be larger than diskquota_naptime?

no, I mean something like https://github.com/RekGRpth/pg_task/blob/7262c04e642ec744d4a34ae3942f9188bfe38409/work.c#L959 where work_sleep function called not frequently than current_timeout irrespective if WaitEventSetWaitMy function returns frequently

from diskquota.

Get your point, but in this bgworker, the wait latch function only receives WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH. I think it will be not frequently triggered by signals.

from diskquota.

I think it will be not frequently triggered by signals.

config reload signal causes WL_LATCH_SET event for example

from diskquota.

I think the configuration reload is not so frequent. And diskquota refresh procedure just updates the size information of tables that become active tables during two refresh procedures. After the current refresh procedure, the next procedure will do fewer updates. Do you have any other example for this scene where the latch will be set several times every minute?

from diskquota.

Do you have any other example for this scene where the latch will be set several times every minute?

this is only theoretical reasoning, what needs to be checked how much time has actually passed since the previous launch

from diskquota.

We will raise a PR to rearrange the bgworker process.

from diskquota.