Re: Startup PANIC on standby promotion due to zero-filled WAL segment
| От | Alena Vinter |
|---|---|
| Тема | Re: Startup PANIC on standby promotion due to zero-filled WAL segment |
| Дата | |
| Msg-id | CAGWv16+kR3OZfcJG8OdoU4Ja5NqqNb6=HM5yGV0zY+yJ0vMuNQ@mail.gmail.com обсуждение исходный текст |
| Ответ на | Re: Startup PANIC on standby promotion due to zero-filled WAL segment (Michael Paquier <michael@paquier.xyz>) |
| Ответы |
Re: Startup PANIC on standby promotion due to zero-filled WAL segment
|
| Список | pgsql-hackers |
> Don't think so, based on what is in the tree:
> $ git grep "fsync = " -- *.pm
> src/test/perl/PostgreSQL/Test/Cluster.pm: print $conf "fsync = off\n";
Oh, didn’t know it's in the `init` function — I apologize.
> The following addition to your proposed test is telling me an entirely
> different story, making the test pass as the records of TLI 1 are
> around:
> my $node_primary = PostgreSQL::Test::Cluster->new('primary');
> $node_primary->init(allows_streaming => 1);
> +#$node_primary->append_conf('postgresql.conf', 'fsync=on');
> $node_primary->start;
I've tried this way, and yes, this works fine. Now I'm really interested in how this parameter prevents the scenario with startup panic. Thank you very much!!!
But I’m still unclear why the segment isn’t copied during replication, as it is in crash recovery (I’d prefer uniform behavior across both paths). Could you help me figure out the answer to that question?
---
> $ git grep "fsync = " -- *.pm
> src/test/perl/PostgreSQL/Test/Cluster.pm: print $conf "fsync = off\n";
Oh, didn’t know it's in the `init` function — I apologize.
> The following addition to your proposed test is telling me an entirely
> different story, making the test pass as the records of TLI 1 are
> around:
> my $node_primary = PostgreSQL::Test::Cluster->new('primary');
> $node_primary->init(allows_streaming => 1);
> +#$node_primary->append_conf('postgresql.conf', 'fsync=on');
> $node_primary->start;
I've tried this way, and yes, this works fine. Now I'm really interested in how this parameter prevents the scenario with startup panic. Thank you very much!!!
But I’m still unclear why the segment isn’t copied during replication, as it is in crash recovery (I’d prefer uniform behavior across both paths). Could you help me figure out the answer to that question?
---
Alena Vinter
В списке pgsql-hackers по дате отправления: