BUG #3731: ash table "PROCLOCK hash" corrupted

"Daniel Cristian Cruz" <cruz@senai-sc.ind.br> writes:

A few moments ago I got the following message, but didn't found any
reference on the internet (including lists).

Nov 8 13:50:56 SERVER postgres[18874]: [5-1] user=XXX,db=XXXPANIC: hash
table "PROCLOCK hash" corrupted

The error showed after some query that where logged because were slow (more
than 5 seconds), in which they had a very big list of values in a NOT IN
clause (not sure if this caused the problem).

I don't know if this is a bug or not.

Well, it shouldn't have happened, so either it's a bug or you had a
hardware glitch. But unless you can find a way to reproduce it I'm
not sure we can do much about it. I doubt your large NOT IN was
relevant --- more likely it'd be something associated with inter-process
interactions.

regards, tom lane

Daniel Cristian Cruz wrote:

A few moments ago I got the following message, but didn't found any
reference on the internet (including lists).

Nov 8 13:50:56 SERVER postgres[18874]: [5-1] user=XXX,db=XXXPANIC: hash
table "PROCLOCK hash" corrupted

The error showed after some query that where logged because were slow (more
than 5 seconds), in which they had a very big list of values in a NOT IN
clause (not sure if this caused the problem).

Can you reproduce it? Have you seen it before?

I don't know if this is a bug or not.

---------------------------(end of broadcast)---------------------------
TIP 6: explain analyze is your friend

--
Alvaro Herrera http://www.CommandPrompt.com/
PostgreSQL Replication, Consulting, Custom Development, 24x7 support

Em Qui, 2007-11-08 às 12:29 -0500, Tom Lane escreveu:

"Daniel Cristian Cruz" <cruz@senai-sc.ind.br> writes:

A few moments ago I got the following message, but didn't found any
reference on the internet (including lists).

Nov 8 13:50:56 SERVER postgres[18874]: [5-1] user=XXX,db=XXXPANIC: hash
table "PROCLOCK hash" corrupted

The error showed after some query that where logged because were slow (more
than 5 seconds), in which they had a very big list of values in a NOT IN
clause (not sure if this caused the problem).

I don't know if this is a bug or not.

Well, it shouldn't have happened, so either it's a bug or you had a
hardware glitch. But unless you can find a way to reproduce it I'm
not sure we can do much about it. I doubt your large NOT IN was
relevant --- more likely it'd be something associated with inter-process
interactions.

It was the first time I saw the message. Never seen it again. I've tried
to reproduce, but it didn't worked.

--
Daniel Cristian Cruz
Analista de Sistemas - Administrador de Banco de Dados
SENAI/SC - Servico Nacional de Aprendizagem Industrial
NTI - Núcleo de Tecnologia da Informação
Fone: (48) 3239-1422