RE: [PERFORM] Dbt2 with postgres issues on CentOS-5.3‏

From: MUHAMMAD ASIF <anaeem(dot)it(at)hotmail(dot)com>
To: <markwkm(at)gmail(dot)com>
Cc: <pgsql-performance(at)postgresql(dot)org>
Subject: RE: [PERFORM] Dbt2 with postgres issues on CentOS-5.3‏
Date: 2010-06-14 18:28:01
Message-ID: BAY154-w577ED84EDB41A1D3FB92C2FFDC0@phx.gbl
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-performance


I am facing sar related issues on Redhat Enterprise Linux64 5.4 too (60G Ram, No Swap space, Xeon Processor).

sar -o /var/dbt2_data/PG/Output/driver/dbt2-sys1/sar_raw.out 60 204
|___ sadc 60 205 -z /var/dbt2_data/PG/Output/driver/dbt2-sys1/sar_raw.out

It generates following sar data i.e.
….
….
03:52:43 AM 2.31
03:53:43 AM 2.31
03:54:43 AM 2.28
03:55:43 AM 2.31
03:56:43 AM 1.67
03:57:43 AM 0.29
03:58:43 AM 0.29
04:00:43 AM 0.30
04:04:00 AM 3.52
04:07:07 AM 0.30
04:09:36 AM 0.23
04:12:04 AM 0.36
04:14:25 AM 0.23
04:16:45 AM 0.26
04:19:10 AM 0.24
04:21:30 AM 0.38
04:23:55 AM 0.24
04:26:25 AM 0.35
04:28:48 AM 0.24
04:31:10 AM 0.27
04:33:40 AM 0.33
04:36:45 AM 0.41
04:39:12 AM 0.27
04:41:41 AM 0.26
04:44:11 AM 0.33
04:46:35 AM 0.25
04:49:06 AM 0.33
04:51:27 AM 0.27
04:53:56 AM 0.23
04:56:19 AM 0.36
04:58:43 AM 0.24
05:01:10 AM 0.35
05:03:43 AM 0.33
05:06:53 AM 0.29
05:09:25 AM 0.23
….
….

To fix this issue I have modified the sysstat-9.1.2/sadc.c and replaced signal based pause (That is not real time) with "select" based pause. That fixed the issue. Thanks.

-------------------------------------------------------------------------------------------------------------------------------------------------------

sadc.c.patch
-------------------------------------------------------------------------------------------------------------------------------------------------------
--- sadc.c.org 2010-06-14 21:44:18.000000000 +0500
+++ sadc.c 2010-06-14 22:52:51.693211184 +0500
@@ -33,6 +33,10 @@
#include <sys/stat.h>
#include <sys/utsname.h>

+#include <sys/types.h>
+#include <sys/time.h>
+#include <time.h>
+
#include "version.h"
#include "sa.h"
#include "rd_stats.h"
@@ -792,6 +796,15 @@
}
}

+void pause_new( void )
+{
+ struct timeval tvsel;
+ tvsel.tv_sec = interval;
+ tvsel.tv_usec = 0;
+
+ select( 0, NULL, NULL, NULL, &tvsel );
+}
+
/*
***************************************************************************
* Main loop: Read stats from the relevant sources and display them.
@@ -899,7 +912,7 @@
}

if (count) {
- pause();
+ pause_new();
}

/* Rotate activity file if necessary */
-------------------------------------------------------------------------------------------------------------------------------------------------------

Best Regards,
Asif Naeem

> Date: Wed, 21 Apr 2010 18:10:35 -0700
> Subject: Re: [PERFORM] Dbt2 with postgres issues on CentOS-5.3‏
> From: markwkm(at)gmail(dot)com
> To: anaeem(dot)it(at)hotmail(dot)com
> CC: pgsql-performance(at)postgresql(dot)org
>
> 2010/4/20 MUHAMMAD ASIF <anaeem(dot)it(at)hotmail(dot)com>:
> > Hi,
> >
> > I am using dbt2 on Linux 64 (CentOS release 5.3 (Final)) . I have compiled
> > latest postgresql-8.4.3 code on the machine and run dbt2 against it. I am
> > little confused about the results. I ran dbt2 with the following
> > configuration i.e.
> >
> > DBT2 Options :
> > WAREHOUSES=75
> > DB_CONNECTIONS=20
> > REGRESS_DURATION=1 #HOURS
> > REGRESS_DURATION_SEC=$((60*60*$REGRESS_DURATION))
> >
> > DBT2 Command :
> > ./dbt2-pgsql-create-db
> > ./dbt2-pgsql-build-db -d $DBDATA -g -r -w $WAREHOUSES
> > ./dbt2-run-workload -a pgsql -c $DB_CONNECTIONS -d
> > $REGRESS_DURATION_SEC -w $WAREHOUSES -o $OUTPUT_DIR
> > ./dbt2-pgsql-stop-db
> >
> > I am not able to understand the sar related graphs. Iostat,mpstat and vmstat
> > results are similar but
> > sar results are strange. I tried to explore the dbt2 source code to find
> > out the how graphs are drawn and why sar results differ.DBT2.pm : 189 reads
> > sar.out and parse it and consider 1 minute elapsed time between each record
> > i.e.
>
> That is certainly a weakness in the logic of the perl modules in
> plotting the charts accurately. I wouldn't be surprised if the other
> stat tools suffer the same problem.
>
> Regards,
> Mark

_________________________________________________________________
Your E-mail and More On-the-Go. Get Windows Live Hotmail Free.
https://signup.live.com/signup.aspx?id=60969

In response to

Browse pgsql-performance by date

  From Date Subject
Next Message Tom Wilcox 2010-06-14 18:53:01 Re: requested shared memory size overflows size_t
Previous Message Lacey Powers 2010-06-14 16:06:55 Re: Re: Command Prompt 8.4.4 PRMs compiled with debug/assert enabled