Re: Database stats ( pg_stat_database.stats_reset ) get reset on daily basis - why?

[Steeve Boulanger <sboulanger29@xxxxxxxxx>]

Apologies for the late reply.

> I'm guessing shutting down Netdata did not stop the resets?

There was a hiccup yesterday, so I had to redo the test and wait until today to see if there were any stats resets ... and NONE today!!

    reset_status    | cnt
--------------------+-----
 reset before today | 100
 no reset           |  57
 reset today        |   0

This is quite interesting, and a little unexpected... but in order to confirm this theory, I have done the following just now: 

1) Restart Netdata service
2) rename the "pg_stat_reset()" function , as you suggested in another post  - very clever btw! :-)  

Now let's wait until tomorrow to see if we have any stats reset and/or some relevant errors in the PG logs! 

Quick note: I did do a quick check in any user functions for any "pg_stat_reset" calls , but didn't find any. With that said, we have 150+ databases in our cluster, so perhaps I missed something. I also did a grep on the netdata directories, but nothing as well. 

-Steeve

On Sat, Nov 23, 2024 at 10:19 AM Adrian Klaver <adrian.klaver@xxxxxxxxxxx> wrote:

On 11/23/24 05:16, Steeve Boulanger wrote:
>> Here (Ireland) we sometimes say "common-or-garden variety".... It means a normal, everyday variety. :-)
>
> I'm afraid that my Irish dialect is limited to "sláinte" only ;-)  In
> any case, thanks for taking the time to help with this issue. I'm
> still investigating, but I think that calling the "ghostbusters" is
> moving up the list now lol.

I'm guessing shutting down Netdata did not stop the resets?

You might try setting log_min_messages to info to see if that catches
anything.


>
>
> On Sat, Nov 23, 2024 at 7:09 AM Ray O'Donnell <ray@xxxxxxxxxxxx> wrote:
>>
>> On 23/11/2024 13:06, Steeve Boulanger wrote:
>>
>>


--
Adrian Klaver
adrian.klaver@xxxxxxxxxxx