Growing Process Lists -

Growing Process Lists -

Post by Andrew Kell » Sun, 18 Aug 2002 02:01:06



I have a new install of Linux (SuSE 7.3) - on two machines - over time
there is a gradual increase in the number of cron sa2 (from SAR), and
sleep processes, seems like two form a day - over time they use
memory, and a reboot is required to keep the system stable ...

What is the source of these processes, and how would I disable them...
A partial dump of ps -ef shows.....

root      1210   778  0 Aug15 ?        00:00:00 /USR/SBIN/CRON
root      1214  1210  0 Aug15 ?        00:00:00 /bin/sh
/usr/lib/sa/sa2 -A
root     23325   778  0 00:00 ?        00:00:00 /USR/SBIN/CRON
root     23329 23325  0 00:00 ?        00:00:00 /bin/sh
/usr/lib/sa/sa2 -A
root      2200   778  0 06:00 ?        00:00:00 /USR/SBIN/CRON
root      2204  2200  0 06:00 ?        00:00:00 /bin/sh
/usr/lib/sa/sa2 -A
root      2631   778  0 12:00 ?        00:00:00 /USR/SBIN/CRON
root      2635  2631  0 12:00 ?        00:00:00 /bin/sh
/usr/lib/sa/sa2 -A
root      2736   363  0 12:00 ?        00:00:00 /usr/sbin/sshd
root      2830  2635  0 12:00 ?        00:00:00 sleep 1
root      2831  2204  0 12:00 ?        00:00:00 sleep 1
root      2832 23329  0 12:00 ?        00:00:00 sleep 1
root      2834  1214  0 12:00 ?        00:00:00 sleep 1

In 40 days this list would be 10 times this size - help

 
 
 

Growing Process Lists -

Post by Jeroen Geilma » Sun, 18 Aug 2002 04:26:37



Quote:> I have a new install of Linux (SuSE 7.3) - on two machines - over time
> there is a gradual increase in the number of cron sa2 (from SAR), and
> sleep processes, seems like two form a day - over time they use
> memory, and a reboot is required to keep the system stable ...

> What is the source of these processes, and how would I disable them...

No. Logs.
Logs logs logs logs - and more logs.

/var/log/sa/*, var/log/messages, /var/log/cron

Are there errors  in any of these ?

To just disable the sar job, delete it from cron (/etc/cron.d/sysstat on
RH7.2 - I don't know where SuSE puts it.)
Alternatively, chmod a-x to clear the executable bits...

Looking at your ps output, it would seem that the sa commands cannot
complete - that's why they are sent to sleep.
Maybe your /var partition is full ? or the write permissions are wrong ?
what's the system load while this happens ?

Research, reasearch...

HTH

 
 
 

Growing Process Lists -

Post by Andrew Kell » Sun, 18 Aug 2002 21:53:21


Excellent starts ... well df shows 37% of the disk in use, space
should not be an issue, /var/log is owned by root, as are the
subdirectories.

A dump of messages looks like this...

linuxbox:/var/log # tail -30 messages
Aug 17 05:59:00 linuxbox /USR/SBIN/CRON[25614]: (root) CMD ( rm -f
/var/spool/cron/lastrun/cron.hourly)
Aug 17 06:00:00 linuxbox /USR/SBIN/CRON[29252]: (root) CMD (
/usr/lib/sa/sa2 -A   #update reports every 6 hour)
Aug 17 06:00:00 linuxbox /USR/SBIN/CRON[29256]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 06:10:00 linuxbox /USR/SBIN/CRON[1396]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 06:20:00 linuxbox /USR/SBIN/CRON[5911]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 06:30:00 linuxbox /USR/SBIN/CRON[10461]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 06:40:00 linuxbox /USR/SBIN/CRON[15021]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 06:50:00 linuxbox /USR/SBIN/CRON[19590]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 06:59:00 linuxbox /USR/SBIN/CRON[20449]: (root) CMD ( rm -f
/var/spool/cron/lastrun/cron.hourly)
Aug 17 07:00:00 linuxbox /USR/SBIN/CRON[24145]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 07:10:00 linuxbox /USR/SBIN/CRON[28712]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 07:20:00 linuxbox /USR/SBIN/CRON[850]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 07:30:00 linuxbox /USR/SBIN/CRON[5421]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 07:40:00 linuxbox /USR/SBIN/CRON[9979]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 07:50:00 linuxbox /USR/SBIN/CRON[14546]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 07:59:01 linuxbox /USR/SBIN/CRON[15408]: (root) CMD ( rm -f
/var/spool/cron/lastrun/cron.hourly)
Aug 17 08:00:00 linuxbox /USR/SBIN/CRON[19040]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 08:10:00 linuxbox /USR/SBIN/CRON[23605]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 08:20:00 linuxbox /USR/SBIN/CRON[28179]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 08:30:00 linuxbox /USR/SBIN/CRON[32743]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 08:40:00 linuxbox /USR/SBIN/CRON[4877]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 08:50:00 linuxbox /USR/SBIN/CRON[9443]: (root) CMD (
/usr/lib/sa/sa1      )
Aug 17 08:59:00 linuxbox /USR/SBIN/CRON[10298]: (root) CMD ( rm -f
/var/spool/cron/lastrun/cron.hourly)

SO - cron is running SA often

and finaly sysstat has the following text....

#crontab for sysstat

#activity reports every 10 minutes everyday
*/10 *   * * *     root  /usr/lib/sa/sa1

#update reports every 6 hour
0    */6 * * *     root  /usr/lib/sa/sa2 -A   #update reports every 6
hour

To which I say I should kill the sysstat entries and that should take
care of it... ? agree

There is nothing in /var/log/sa that is readable




> > I have a new install of Linux (SuSE 7.3) - on two machines - over time
> > there is a gradual increase in the number of cron sa2 (from SAR), and
> > sleep processes, seems like two form a day - over time they use
> > memory, and a reboot is required to keep the system stable ...

> > What is the source of these processes, and how would I disable them...

> No. Logs.
> Logs logs logs logs - and more logs.

> /var/log/sa/*, var/log/messages, /var/log/cron

> Are there errors  in any of these ?

> To just disable the sar job, delete it from cron (/etc/cron.d/sysstat on
> RH7.2 - I don't know where SuSE puts it.)
> Alternatively, chmod a-x to clear the executable bits...

> Looking at your ps output, it would seem that the sa commands cannot
> complete - that's why they are sent to sleep.
> Maybe your /var partition is full ? or the write permissions are wrong ?
> what's the system load while this happens ?

> Research, reasearch...

> HTH

 
 
 

Growing Process Lists -

Post by Jeroen Geilma » Mon, 19 Aug 2002 00:06:44



Quote:> Excellent starts ... well df shows 37% of the disk in use, space
> should not be an issue, /var/log is owned by root, as are the
> subdirectories.

This is not at all true for all log directories....(again, on my RH7.2 box)

Quote:

> A dump of messages looks like this...

<snippety snip>

Quote:> SO - cron is running SA often

Yes, it does - it takes a snapshot of running processes, system load and
memory usage.

Quote:

> and finaly sysstat has the following text....

> #crontab for sysstat

> #activity reports every 10 minutes everyday
> */10 *   * * *     root  /usr/lib/sa/sa1

> #update reports every 6 hour
> 0    */6 * * *     root  /usr/lib/sa/sa2 -A   #update reports every 6
> hour

> To which I say I should kill the sysstat entries and that should take
> care of it... ? agree

Yes, it should - but that still doesn't explain why it fails to run
properly.

Quote:

> There is nothing in /var/log/sa that is readable

No, it's a binary report meant to be interpreted by specialist log
analysers.

I never managed to figure out what to do with them, anyway ;-)

Just disable it.

 
 
 

Growing Process Lists -

Post by Andrew Kell » Mon, 19 Aug 2002 11:38:24


Right - it is disabled -
Right again - the problem is that sar is not running correctly - given
it starts every 10 minutes and 2-3 processes occur a day that is not
too bad, but I am not going to waste my time debuging it's problem -
will remove from crontab

thanks for the input.

amk

 
 
 

1. RedHat 6.0: rpc.mountd process grows and grows.

We have a RedHat 6.0 machine (kernel 2.2.9) with 256MB RAM acting as an NFS
server, with moderate use from various NFS clients (Linux RH 6.0 and
NEXTSTEP primarily).  After not too much time (say a day or two), the
rpc.mountd process grows to be VERY large, here's an example of an
rpc.mountd's process information (from top)after running for just 1 day:
PID    USER    PRI     NI    SIZE    RSS    SHARE    STAT    LIB    %CPU
%MEM    TIME    COMMAND
xxxx   root    4       0     159M    143M   433        S     0      0.0
56.8    3:40    rpc.mountd

Eventually, this process grows and grows and eventually dies.  At that time,
I have to restart the daemon.  No apparent useful information seems to be
logged to /var/log/messages.

Is this expected behavior?  Can anything be done?

--
Rex Dieter
Computer System Administrator
Mathematics and Statistics
University of Nebraska Lincoln

2. TCP/IP network programming guide

3. syslogd grows and grows and grows on 2.6/x86

4. Disconnecting Users on demand

5. why does /var/adm/wtmp grow and grow and grow ...

6. linux boot problem

7. "Cached" grows and grows and grows...

8. bbs for linux???

9. Apache growing,Growing,GROWING in memory usage

10. startx problem. The list of people grows.

11. /usr/lpd/qadm.config grows and grows

12. root directory growing and growing and ....

13. FWD: SuSE and TurboLinux Join Growing List of Linux Java Technology Distributors.