UPDATE stalls when run in "batch mode" - Mailing list pgsql-general

From Kynn Jones
Subject UPDATE stalls when run in "batch mode"
Date
Msg-id c2350ba40803161305m41479987la5c4ddabea33c87b@mail.gmail.com
Whole thread Raw
Responses Re: UPDATE stalls when run in "batch mode"  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-general




I was running an SQL file in psql (via \i) and I noticed that the execution had been stuck at a particular place for a few hours, which was far longer than expected.

So I killed the processing of the file (with Ctrl-C), vacuumed everything I could think of and tried again.

The same thing happened: the processing stalled at the same place as before.

Once more I killed the processing of the file, but then, this time, instead of re-running the whole thing from the beginning like I did before, I just attempted to execute the statement at which the processing was hanging.  It was in fact an update statement of the form:

UPDATE T SET K = NULL FROM B WHERE T.K = B.X;

For good measure, I then executed the next few statements individually, and they all finished in a few seconds.  Then I ran the remainder of the original file, and everything went without a hitch, at the expected speed.

I can't begin to guess why the update statement above caused the processing to stall, but ran quickly when I executed it by itself.  How can I understand this situation better?  I.e. is there any diagnostic procedure I can perform that may lead to the explanation for this inconsistent behavior?

(FWIW, none of the statements in question were executed within a transaction block.)

TIA!

Kynn

pgsql-general by date:

Previous
From: "Scott Marlowe"
Date:
Subject: Re: Redundant file server for postgres
Next
From: Robert Treat
Date:
Subject: Re: Trigger to run @ connection time?