List:General Discussion« Previous MessageNext Message »
From:Bernd Jagla Date:March 2 2004 7:37pm
Subject:Re: no one can log on any more
View as plain text  
table: text_data
type: ALL
possible_keys: NULL
key: NULL
key_len: NULL
ref: NULL
rows 133856002
Extra: Using where

No temp tables created, no i/o problem, no swapping.


B
----- Original Message ----- 
From: <vpendleton@stripped>
To: "Bernd Jagla" <jaglab@stripped>
Cc: "mysql" <mysql@stripped>
Sent: Tuesday, March 02, 2004 2:34 PM
Subject: Re: no one can log on any more


Can you post the explain plan? I would be interested to see if temp table
tables are being created or if the system is swapping or if the the temp
tables are so large that the entire system begins to act poorly.

>>>>>>>>>>>>>>>>>> Original Message
> <<<<<<<<<<<<<<<<<<

On 3/2/04, 12:35:20 PM, Bernd Jagla <jaglab@stripped> wrote regarding Re:
no one can log on any more:


> There are no indices on the new table.
> Top looks normal: approx 100-150% usage (out of 800% [8 cpus])
> There is no major i/o problem: sar -d says that less than 2% of io are
used.

> The only problems in this direction might be connected with memory and
the
> associated parameters in mysql.
> The file system cache is 5GB. I hope this is no problem. mysqld uses
about
> 54619 K (gmemusage)
> Top shows the actual size as 1794M and Res=1609

> The problem only occurs when I do two similar queries at the same time.
Both
> queries access the same table for read (not write). Might this be the
> problem? How would I solve this one?

> Bernd

> ----- Original Message -----
> From: <vpendleton@stripped>
> To: "Bernd Jagla" <jaglab@stripped>
> Cc: "mysql" <mysql@stripped>
> Sent: Tuesday, March 02, 2004 1:17 PM
> Subject: Re: no one can log on any more


> When the queries in question are running, what does top look like? With
> an insert that large it may be beneficial to disable any keys that may be
> active on the target table and re-enable them after the data has been
> loaded.

> >>>>>>>>>>>>>>>>>> Original
> Message <<<<<<<<<<<<<<<<<<

> On 3/2/04, 12:08:05 PM, Bernd Jagla <jaglab@stripped> wrote regarding Re:
> no one can log on any more:


> > An explain on the select statements reveals that no index is used and
> that
> > all 133856002 rows are searched.

> > Do you think that using a new index would make much of difference?
> > I think it has something to do with some system/mysql parameters that
can
> be
> > adjusted.
> > I also don't understand why no one else can log on to the mysql system.

> > If you or someone could explain or point me to a documentation on this
it
> > would be pretty close to what I want.

> > Thanks for your help

> > Bernd

> > ----- Original Message -----
> > From: <vpendleton@stripped>
> > To: "Bernd Jagla" <jaglab@stripped>
> > Cc: "mysql" <mysql@stripped>
> > Sent: Tuesday, March 02, 2004 1:02 PM
> > Subject: Re: no one can log on any more


> > What does the explain plan reveal?

> > >>>>>>>>>>>>>>>>>>
> Original Message <<<<<<<<<<<<<<<<<<

> > On 3/2/04, 9:05:17 AM, Bernd Jagla <jaglab@stripped> wrote regarding no
> > one can log on any more:


> > > Hi there,

> > > I have some weird sql statements that seem to eat up all resources
from
> > > mysqld.
> > > Once I have two of them running at the same time no one can log on any
> > > more. Simple queries involving distinct get a out of memory error.

> > > We have tried changing the index buffer size from 256M to 2G, no
change.
> > > We have 7G of memory on an IRIX system with 8 cpus.
> > > No replication.

> > > Do you have any idea what is happening?

> > > Thanks a lot

> > > Bernd


> > > the sql statements look like this:

> > > insert into rule2
> > > select id
> > > from text_uniq_bin
> > > where (substring(rev_rep,1,1)="1") AND
> > > (
> > > substring(replaced,1,1) +
> > > substring(replaced,2,1) +
> > > substring(replaced,3,1) +
> > > substring(replaced,4,1) +
> > > substring(replaced,5,1) +
> > > substring(replaced,6,1) +
> > > substring(replaced,7,1) +
> > > substring(replaced,8,1) +
> > > substring(replaced,9,1)
> > > ) >2 AND
> > > (substring(replaced,3,1)="0") AND
> > > (
> > > substring(replaced,12,1) +
> > > substring(replaced,13,1) +
> > > substring(replaced,14,1) +
> > > substring(replaced,15,1) +
> > > substring(replaced,16,1) +
> > > substring(replaced,17,1) +
> > > substring(replaced,18,1) +
> > > substring(replaced,19,1)
> > > ) >5
> > > ;

> > > AND

> > > insert into rule3
> > > select id
> > > from text_uniq_bin
> > > where (substring(rev_rep,1,1)="1") AND
> > > (
> > > substring(replaced,1,1) +
> > > substring(replaced,2,1) +
> > > substring(replaced,3,1) +
> > > substring(replaced,4,1) +
> > > substring(replaced,5,1) +
> > > substring(replaced,6,1) +
> > > substring(replaced,7,1) +
> > > substring(replaced,8,1) +
> > > substring(replaced,9,1)
> > > ) >2 AND
> > > (substring(rev_rep,3,1)="0") AND
> > > (
> > > substring(replaced,12,1) +
> > > substring(replaced,13,1) +
> > > substring(replaced,14,1) +
> > > substring(replaced,15,1) +
> > > substring(replaced,16,1) +
> > > substring(replaced,17,1) +
> > > substring(replaced,18,1) +
> > > substring(replaced,19,1)
> > > ) <=5 AND
> > > (substring(rev_rep,2,1)="0") AND
> > > (
> > > substring(replaced,6,1) +
> > > substring(replaced,7,1) +
> > > substring(replaced,8,1) +
> > > substring(replaced,9,1) +
> > > substring(replaced,10,1) +
> > > substring(replaced,11,1) +
> > > substring(replaced,12,1) +
> > > substring(replaced,13,1) +
> > > substring(replaced,14,1) +
> > > substring(replaced,15,1) +
> > > substring(replaced,16,1) +
> > > substring(replaced,17,1) +
> > > substring(replaced,18,1) +
> > > substring(replaced,19,1)
> > > ) >7


Thread
no one can log on any moreBernd Jagla2 Mar
  • Re: no one can log on any morevpendleton2 Mar
  • Re: no one can log on any moreBernd Jagla2 Mar
    • Re: no one can log on any morevpendleton2 Mar
  • Re: no one can log on any moreBernd Jagla2 Mar
    • Re: no one can log on any morevpendleton2 Mar
  • Re: no one can log on any moreBernd Jagla2 Mar
    • Re: no one can log on any morevpendleton2 Mar
  • Re: no one can log on any moreBernd Jagla2 Mar
    • Re: no one can log on any morevpendleton2 Mar
  • Re: no one can log on any moreBernd Jagla2 Mar
    • Re: no one can log on any morevpendleton2 Mar
  • Re: no one can log on any moreMartijn Tonies2 Mar
  • Re: no one can log on any moreBernd Jagla2 Mar
    • Re: no one can log on any morevpendleton2 Mar
  • Re: no one can log on any moreSasha Pachev2 Mar