Re: [QUESTIONS] Business cases

From: "Igor Sysoev" <igor(at)nitek(dot)ru>
To: <pgsql-questions(at)postgresql(dot)org>
Subject: Re: [QUESTIONS] Business cases
Date: 1998-01-21 13:34:25
Message-ID: 199801211337.QAA24083@ns.nitek.ru
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

> > In all tests crash_me crashed backend server - postgres but postmaster
> > still run and allow to work with all dbs. I run crash_me several times
> > with postmaster started once upon boot time.
>
> Is there any indication as to why the crash?

I can tell indirect only using limits/pg.cfg where crash_me logs all
results.
Every time it runs it checks this file to set limits in the test to avoid
crash.
I think the best way to look PGSQL source code and set up checkups
for these limits. So :

1. output:
query size: Broken pipe

pg.cfg:
query_size=4096 # query size

2. output:
constant string size in where:
Fatal error: Can't check 'constant string size in where' for limit=1
error: PQexec() -- There is no connection to the backend.

pg.cfg:
where_string_size=4062 # constant string size in where

3. output:
tables in join:
Fatal error: Can't check 'tables in join' for limit=1
error: PQexec() -- There is no connection to the backend.

pg.cfg:
join_tables=0 # tables in join

4. output:
table name length:
Fatal error: Can't check 'table name length' for limit=1
error: PQexec() -- There is no connection to the backend.

pg.cfg:
max_table_name=31 # table name length

5. output:
max table row length (without blobs):
Fatal error: Can't check 'max table row length (without blobs)' for limit=1
error: PQexec() -- There is no connection to the backend.

pg.cfg:
max_row_length=7969 # max table row length (without blobs)

When I run tests first time I crashed at "column name length: 31" also.
Now it runs OK - 8-[.

And I cann't indicate couse of the problem with "index length" - it run out
of memory.

Igor Sysoev

Browse pgsql-hackers by date

  From Date Subject
Next Message Karl Denninger 1998-01-21 13:53:02 Re: [QUESTIONS] Re: locking
Previous Message Vadim B. Mikheev 1998-01-21 10:10:22 Re: subselects