Обсуждение: Where do we stand on 9.3 bugs?
What remaining issues are there blocking a 9.3.3 release? I know that
there were unresolved multixact issues when we put out 9.3.2 --- are
those all dealt with now? What else do people see as release-blockers?
regards, tom lane
Tom Lane wrote: > What remaining issues are there blocking a 9.3.3 release? I know that > there were unresolved multixact issues when we put out 9.3.2 --- are > those all dealt with now? What else do people see as release-blockers? The only thing I'm aware still outstanding in multixact land is the creation of a separate multixact freeze age GUC variable. All other issues were resolved. Bug #8470 is still unresolved, but that's a performance issue. -- Álvaro Herrera http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services
On 2014-01-13 12:26:45 -0300, Alvaro Herrera wrote: > Tom Lane wrote: > > What remaining issues are there blocking a 9.3.3 release? I know that > > there were unresolved multixact issues when we put out 9.3.2 --- are > > those all dealt with now? What else do people see as release-blockers? > > The only thing I'm aware still outstanding in multixact land is the > creation of a separate multixact freeze age GUC variable. All other > issues were resolved. Bug #8470 is still unresolved, but that's a > performance issue. It's not strictly a 9.3 issue, but there's still the issue of /members wrapping around into valid data, right? Greetings, Andres Freund -- Andres Freund http://www.2ndQuadrant.com/PostgreSQL Development, 24x7 Support, Training & Services
Andres Freund wrote: > On 2014-01-13 12:26:45 -0300, Alvaro Herrera wrote: > > Tom Lane wrote: > > > What remaining issues are there blocking a 9.3.3 release? I know that > > > there were unresolved multixact issues when we put out 9.3.2 --- are > > > those all dealt with now? What else do people see as release-blockers? > > > > The only thing I'm aware still outstanding in multixact land is the > > creation of a separate multixact freeze age GUC variable. All other > > issues were resolved. Bug #8470 is still unresolved, but that's a > > performance issue. > > It's not strictly a 9.3 issue, but there's still the issue of /members > wrapping around into valid data, right? Yes, but TBH I don't have much of an intention to fix that other than having the freeze limit. At least not for 9.3.3. -- Álvaro Herrera http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services
On Mon, Jan 13, 2014 at 5:16 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote: > What remaining issues are there blocking a 9.3.3 release? Well hardly a blocker since this has missed 2 releases already, but I'm still hopeful to get many PGXS-based extensions to build again without the dreaded "install: will not overwrite just-created ..." http://www.postgresql.org/message-id/52406191.6040508@dunslane.net Thankfully I'm mostly using Debian so it's already patched for me. Regards, Marti
On 01/13/2014 07:16 AM, Tom Lane wrote: > What remaining issues are there blocking a 9.3.3 release? I know that > there were unresolved multixact issues when we put out 9.3.2 --- are > those all dealt with now? What else do people see as release-blockers? I see this bug as a release-blocker. It's a fairly crippling replication bug for a bunch of users (5 reports to date): http://archives.postgresql.org/message-id/20140106134815.GE28320%40alap2.anarazel.de Despite actually dating back to 9.0, something in the 9.3.2/9.2.6 updates is causing users to hit it now. And for those who do hit it, replication is impossible and there's no workaround. -- Josh Berkus PostgreSQL Experts Inc. http://pgexperts.com
On 2014-01-13 17:30:55 -0800, Josh Berkus wrote: > On 01/13/2014 07:16 AM, Tom Lane wrote: > > What remaining issues are there blocking a 9.3.3 release? I know that > > there were unresolved multixact issues when we put out 9.3.2 --- are > > those all dealt with now? What else do people see as release-blockers? > > I see this bug as a release-blocker. It's a fairly crippling > replication bug for a bunch of users (5 reports to date): > http://archives.postgresql.org/message-id/20140106134815.GE28320%40alap2.anarazel.de Absolutely. I didn't list it because it's not 9.3 specific. I think Tom is working on a fix right now, if not, Heikki has showed interest and I would also work on it once the CF has started. So I don't think there's danger of missing that one. > Despite actually dating back to 9.0, something in the 9.3.2/9.2.6 > updates is causing users to hit it now. And for those who do hit it, > replication is impossible and there's no workaround. There have been reports about it for years if you search in the archives and the code around it hasn't changed for a fair bit of time, so I think that's just coincidence. Greetings, Andres Freund -- Andres Freund http://www.2ndQuadrant.com/PostgreSQL Development, 24x7 Support, Training & Services
Andres Freund <andres@2ndquadrant.com> writes:
> On 2014-01-13 17:30:55 -0800, Josh Berkus wrote:
>> Despite actually dating back to 9.0, something in the 9.3.2/9.2.6
>> updates is causing users to hit it now. And for those who do hit it,
>> replication is impossible and there's no workaround.
> There have been reports about it for years if you search in the archives
> and the code around it hasn't changed for a fair bit of time, so I think
> that's just coincidence.
I looked at the commit history for 9.2 post-release, and I don't see
anything that looks related, so I also doubt the theory that some recent
change has increased the odds of this happening. I think it's more that
we now recognize the cause of what previously seemed random breakage.
regards, tom lane
Marti Raudsepp <marti@juffo.org> writes: > On Mon, Jan 13, 2014 at 5:16 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote: >> What remaining issues are there blocking a 9.3.3 release? > Well hardly a blocker since this has missed 2 releases already, but > I'm still hopeful to get many PGXS-based extensions to build again > without the dreaded "install: will not overwrite just-created ..." > http://www.postgresql.org/message-id/52406191.6040508@dunslane.net AFAICT, the state of play on that is that it was reverted out of the non-devel branches immediately before the October releases (9.3.1 et al), on the grounds that it still seemed to be a moving target, cf http://www.postgresql.org/message-id/1381193255.25702.4.camel@vanquo.pezone.net I've not seen any further work on PGXS since commit eebdea08b, so maybe that complaint is no longer valid? On the other hand, dddc91ddd leaves the impression that all is not yet well in PGXS-land. > Thankfully I'm mostly using Debian so it's already patched for me. The Debian guys do seem remarkably willing to apply patches that upstream doesn't trust yet, don't they. Anyway, my opinion about this one is it's not a regression and therefore not a blocker, but I have no objection to fixing it in the back branches once we have consensus that the fix is stable. regards, tom lane