Re: PITR and Temp Tables

2022-04-20 Thread Huan Ruan
Many thanks for the explanation, Tom and Greg. That all makes sense. Cheers Huan

Re: PITR and Temp Tables

2022-04-20 Thread Huan Ruan
Hi Patrick Thanks for your reply. > > > My guess is that temp table entries will still be in your catalog until > you do a VACUUM FULL of the pg_class / pg_attribute tables. > > But you should not care about these entries if these tables are vacuumed > at regular intervals. > What I observed in o

PITR and Temp Tables

2022-04-20 Thread Huan Ruan
Hi All Let's say at T0 a database has N session based temp tables. They would have corresponding records in the catalog tables like pg_class and pg_attribute that are visible to other sessions. At T1, I do a PITR to T0. That recovered database should not have those temp tables because the session

Re: Potential BRIN Index Corruption

2020-12-13 Thread Huan Ruan
Hi Tomas The records have a timestamp column so we do know the time they were written. We didn't find any I/O issues that match that time but unfortunately as it's been a while we are not confident with that finding. > Are there any other corrupted indexes on the table? That was one of my first

Re: Potential BRIN Index Corruption

2020-12-08 Thread Huan Ruan
HI Alvaro Unfortunately those records were written a while ago and we no longer keep their WAL logs. Thanks for your help anyway. Huan On Fri, 27 Nov 2020 at 08:40, Huan Ruan wrote: > Thanks Alvaro for pointing me to pageinspect. I will give it a try and >> report back. It might t

Re: Potential BRIN Index Corruption

2020-11-26 Thread Huan Ruan
> > Thanks Alvaro for pointing me to pageinspect. I will give it a try and > report back. It might take a few days. I knew this module but have never > tried it before. >

Potential BRIN Index Corruption

2020-11-25 Thread Huan Ruan
Hi All We have a table with 623 million records. It appears a BRIN index of this table on a timestamp column is missing some records, as illustrated below in a cut-down version with additional columns and indices omitted. We cannot work out a reproducible case but we have a copy of the offending