Re: Toast issues with OldestXmin going backwards - Mailing list pgsql-hackers

From Andrew Gierth
Subject Re: Toast issues with OldestXmin going backwards
Date
Msg-id 87tvrxa210.fsf@news-spur.riddles.org.uk
Whole thread Raw
In response to Re: Toast issues with OldestXmin going backwards  (Robert Haas <robertmhaas@gmail.com>)
List pgsql-hackers
>>>>> "Robert" == Robert Haas <robertmhaas@gmail.com> writes:

 Robert> I wonder if perhaps get_actual_variable_range() has a hazard of
 Robert> this type as well. If OldestXmin retreats, then the special
 Robert> snapshot type which it uses could see a row whose TOAST entry
 Robert> has been removed meanwhile.

Actually get_actual_variable_range may have a hazard even if OldestXmin
does not retreat; it uses RecentGlobalXmin, which is quite likely to be
older than an OldestXmin value used by a vacuum.

-- 
Andrew (irc:RhodiumToad)


pgsql-hackers by date:

Previous
From: Andrew Gierth
Date:
Subject: Re: Toast issues with OldestXmin going backwards
Next
From: Thomas Munro
Date:
Subject: Re: PostgreSQL's handling of fsync() errors is unsafe and risks data loss at least on XFS