Re: [HACKERS] statistics for shared catalogs not updated when autovacuum is off

Noah Misch writes: > You may want to compare your patch to this pending patch for the same bug: > http://www.postgresql.org/message-id/flat/24f09c2d-e5bf-1f73-db54-8255c1280...@2ndquadrant.com Oh, interesting. I had not been paying any attention to that thread. I'll go

Re: [HACKERS] statistics for shared catalogs not updated when autovacuum is off

On Wed, May 25, 2016 at 05:50:41PM -0400, Tom Lane wrote: > I wrote: > > OK. I'm going to work on replacing the DBWriteRequest structure with > > an OID list, and on improving the comments some more, but I don't > > anticipate further functional changes. > > Here's the end result of that. You

Re: [HACKERS] statistics for shared catalogs not updated when autovacuum is off

I wrote: > OK. I'm going to work on replacing the DBWriteRequest structure with > an OID list, and on improving the comments some more, but I don't > anticipate further functional changes. Here's the end result of that. I went ahead and pushed the original small patch, concluding that that was

Re: [HACKERS] statistics for shared catalogs not updated when autovacuum is off

Michael Paquier writes: > On Tue, May 24, 2016 at 3:52 PM, Tom Lane wrote: >> True, although I'm not sure if adding a dependency on IsSharedRelation() >> here is a good thing. In any case, you took the dependency too far ... > OK. It seemed more

Re: [HACKERS] statistics for shared catalogs not updated when autovacuum is off

On Tue, May 24, 2016 at 3:52 PM, Tom Lane wrote: > Michael Paquier writes: >> At the same time I am not getting why pgstat_fetch_stat_tabentry needs >> to be that complicated. Based on the relation OID we can know if it is >> a shared relation or

Re: [HACKERS] statistics for shared catalogs not updated when autovacuum is off

I wrote: > In short, I think the attached is enough to fix it. There's some cosmetic > cleanup that could be done here: a lot of these functions could use better > comments, and I'm not happy about the autovac launcher depending on > MyDatabaseId being zero. But those are not functional

Re: [HACKERS] statistics for shared catalogs not updated when autovacuum is off

Michael Paquier writes: > The problem is different I think. Since 9.3, database-related > statistics are located on separate files. And the statistics of shared > tables is visibly located in a file with database name set as > InvalidOid, leading to the presence of

Re: [HACKERS] statistics for shared catalogs not updated when autovacuum is off

On Tue, Feb 2, 2016 at 10:38 AM, Jim Nasby wrote: > On 2/1/16 7:20 PM, Peter Eisentraut wrote: >> That's probably right. Even with autovacuum on, the statistics for >> shared catalogs do not appear as updated right away. That is, if you >> run VACUUM and then look at

Re: [HACKERS] statistics for shared catalogs not updated when autovacuum is off

On 2/1/16 9:49 AM, Jim Nasby wrote: > On 1/30/16 5:05 PM, Peter Eisentraut wrote: >> When autovacuum is off, the statistics in pg_stat_sys_tables for shared >> catalogs (e.g., pg_authid, pg_database) never update. So seq_scan >> doesn't update when you read the table, last_analyze doesn't update

Re: [HACKERS] statistics for shared catalogs not updated when autovacuum is off

On 2/1/16 7:20 PM, Peter Eisentraut wrote: That's probably right. Even with autovacuum on, the statistics for shared catalogs do not appear as updated right away. That is, if you run VACUUM and then look at pg_stat_sys_tables right away, you will see the stats for shared catalogs to be

Re: [HACKERS] statistics for shared catalogs not updated when autovacuum is off

On 1/30/16 5:05 PM, Peter Eisentraut wrote: When autovacuum is off, the statistics in pg_stat_sys_tables for shared catalogs (e.g., pg_authid, pg_database) never update. So seq_scan doesn't update when you read the table, last_analyze doesn't update when you run analyze, etc. But when you shut

[HACKERS] statistics for shared catalogs not updated when autovacuum is off

When autovacuum is off, the statistics in pg_stat_sys_tables for shared catalogs (e.g., pg_authid, pg_database) never update. So seq_scan doesn't update when you read the table, last_analyze doesn't update when you run analyze, etc. But when you shut down the server and restart it with