Upgrade from 9.1.6 to 9,1.7 seems to have caused a major slowdown

[Greg Williamson]

Dear peoples,


Today we upgraded a small (7.5 megabytes) but hyperactive database from postgresql 9.1.6 to 9.1.7 (we use INDEX CONCURRENTLY a lot) and after restarting it looks as if we have a slower system -- fewer queries, but more long ones, fewer checkpoints. I am still manually comparing config  settings to see if anything changed (shouldn't have but ...).

Has anyone else seen anything like this ?

I  suspect some site error had led to us scoring an own-goal but I thought I'd ask and see if anyone else has seen this.


Thanks,

Greg Williamson

[Rafael Martinez]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Post by Greg Williamson
Dear peoples,
Today we upgraded a small (7.5 megabytes) but hyperactive database
from postgresql 9.1.6 to 9.1.7 (we use INDEX CONCURRENTLY a lot)
and after restarting it looks as if we have a slower system --
fewer queries, but more long ones, fewer checkpoints. I am still
manually comparing config settings to see if anything changed
(shouldn't have but ...).
Has anyone else seen anything like this ?

Hello

The release notes for 9.2.2 say this:

"... However, you may need to perform REINDEX operations to correct
problems in concurrently-built indexes, as described in the first
changelog item below....."

But the 9.1.7 note says nothing about this, even when this version has
the same patch for 'CREATE INDEX CONCURRENTLY' as 9.2.2

Why don't you try to reindex the indexes in your database and see it
this helps?

regards,
- --
Rafael Martinez Guerrero
Center for Information Technology
University of Oslo, Norway

PGP Public Key: http://folk.uio.no/rafael/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.14 (GNU/Linux)

iEYEARECAAYFAlDTGH0ACgkQBhuKQurGihSLbQCeI1J49iLrlGcYJmiWCUnK5g8Y
Ca8AnA06l9wEa2fLNRKpzktcgL4AkKBk
=ygjT
-----END PGP SIGNATURE-----

[Greg Williamson]

Rafael --
<...>

Post by Rafael Martinez

Post by Greg Williamson
Has anyone else seen anything like this ?

Hello
"... However, you may need to perform REINDEX operations to correct
problems in concurrently-built indexes, as described in the first
changelog item below....."
But the 9.1.7 note says nothing about this, even when this version has
the same patch for 'CREATE INDEX CONCURRENTLY' as 9.2.2
Why don't you try to reindex the indexes in your database and see it
this helps?
regards,
- --
Rafael Martinez Guerrero

Thanks for the suggestion -- I did indeed a lot of time on checking indexes.

But of course it was not postgres at all, just servers that hadn't dealt well with the brief outage.

The upgraded instance is performing just as expected.

Thanks & sorry for wasting time of the list, wee hours of the morning and a sense of desperation.


GW