Re: Why doesn't GiST VACUUM require a super-exclusive lock, like nbtree VACUUM? - Mailing list pgsql-hackers

From Michail Nikolaev
Subject Re: Why doesn't GiST VACUUM require a super-exclusive lock, like nbtree VACUUM?
Date
Msg-id CANtu0ojcU0TVZwMYNvJ85615KqVoLaJVDouWaJL54FofxHDUdw@mail.gmail.com
Whole thread Raw
In response to Why doesn't GiST VACUUM require a super-exclusive lock, like nbtree VACUUM?  (Peter Geoghegan <pg@bowt.ie>)
Responses Re: Why doesn't GiST VACUUM require a super-exclusive lock, like nbtree VACUUM?
List pgsql-hackers
Hello.

One thing I think we could add to the patches is to adapt the 10-years-old comment below with notice about IOS:

/*
* We save the LSN of the page as we read it, so that we know whether it
* safe to apply LP_DEAD hints to the page later. This allows us to drop
* the pin for MVCC scans, which allows vacuum to avoid blocking.
*/
so->curPageLSN = BufferGetLSNAtomic(buffer);


Also, I think it is a good idea to add "Assert(!scan->xs_want_itup);" to gistkillitems.

Best regards,
Mikhail.

pgsql-hackers by date:

Previous
From: Andres Freund
Date:
Subject: Re: AIO v2.0
Next
From: Ilia Evdokimov
Date:
Subject: Re: Sample rate added to pg_stat_statements