Re: Possible performance regression with pg_dump of a large number of relations

Hello Stephen, everyone, On Fri, Jan 12, 2018 at 10:39:09PM -0500, Stephen Frost wrote: > Greetings Jeff & Luke, > > * Jeff Janes (jeff.ja...@gmail.com) wrote: > > Sorry, that query reflects some munging I did to it. The real part added > > to the query is: > > > > EXISTS (SELECT 1 FROM pg_attrib

Re: Possible performance regression with pg_dump of a large number of relations

Luke, * Luke Cowell (lcow...@gmail.com) wrote: > > On Jan 24, 2018, at 2:56 PM, Stephen Frost wrote: > >>> ERROR: relation "pg_init_privs" does not exist > >>> LINE 139: LEFT JOIN pg_init_privs pip > > > > I certainly hope that works on 9.6, since that's when pg_init_privs was > > added.. > >

Re: Possible performance regression with pg_dump of a large number of relations

> On Jan 24, 2018, at 2:56 PM, Stephen Frost wrote: > > Hi there! > > >>> ERROR: relation "pg_init_privs" does not exist >>> LINE 139: LEFT JOIN pg_init_privs pip > > I certainly hope that works on 9.6, since that's when pg_init_privs was > added.. My mistake. That error is from my 9.5 ser

Re: Possible performance regression with pg_dump of a large number of relations

Hi there! * Luke Cowell (lcow...@gmail.com) wrote: > Hi Stephen, thank you for putting this together. Yeah, it needs more work, which I figured out after actually hacking together a patch for it and I've just not gotten back to it yet. > > If folks get a chance to take a look at the query and/or

Re: Possible performance regression with pg_dump of a large number of relations

Hi Stephen, thank you for putting this together. > If folks get a chance to take a look at the query and/or test, that'd be > great. I'll try to work up an actual patch to pg_dump this weekend to > run it through the regression tests and see if anything breaks. I'm not sure how I can help other

Re: Possible performance regression with pg_dump of a large number of relations

Greetings Jeff & Luke, * Jeff Janes (jeff.ja...@gmail.com) wrote: > Sorry, that query reflects some munging I did to it. The real part added > to the query is: > > EXISTS (SELECT 1 FROM pg_attribute at LEFT JOIN pg_init_privs pip ON(c.oid > = pip.objoid AND pip.classoid = (SELECT oid FROM pg_cla

Re: Possible performance regression with pg_dump of a large number of relations

> On Jan 12, 2018, at 8:01 AM, Jeff Janes wrote: > > On Thu, Jan 11, 2018 at 5:26 PM, Luke Cowell > wrote: > I've been troubleshooting an issue with slow pg_dump times on postgres 9.6.6. > I believe something changed between 9.5.10 and 9.6.6 that has made dumps > sign

Re: Possible performance regression with pg_dump of a large number of relations

On Fri, Jan 12, 2018 at 8:01 AM, Jeff Janes wrote: > That commit covered a few different things, and I don't what improvement > it mentions is the one that motivated this, but the key change was to add > this query: > > EXISTS (SELECT 1 FROM pg_attribute at LEFT JOIN pg_init_privs pip ON(c.oid >

Re: Possible performance regression with pg_dump of a large number of relations

On Thu, Jan 11, 2018 at 5:26 PM, Luke Cowell wrote: > I've been troubleshooting an issue with slow pg_dump times on postgres > 9.6.6. I believe something changed between 9.5.10 and 9.6.6 that has made > dumps significantly slower for databases with a large number of relations. > I posted this in

Possible performance regression with pg_dump of a large number of relations

I've been troubleshooting an issue with slow pg_dump times on postgres 9.6.6. I believe something changed between 9.5.10 and 9.6.6 that has made dumps significantly slower for databases with a large number of relations. I posted this in irc and someone suggested that I should post this here. I'm