sar -q and ps aux
 

Hi!

Im my 2 nodes of oracle RAC server production database:

cat /etc/redhat-release
Red Hat Enterprise Linux Server release 5.8 (Tikanga)

When I try to identify how many processes are running I get the following result:

ps aux | wc -l
924

But It is totally different on sar command, look at plist-sz collumn:

1:00:01 AM runq-sz plist-sz ldavg-1 ldavg-5 ldavg-15
11:10:01 AM 1 18091 3.23 3.18 3.45
11:20:01 AM 1 18088 3.47 3.10 3.23
11:30:01 AM 1 18074 3.65 3.35 3.29
11:40:01 AM 2 18068 3.43 3.45 3.34
11:50:01 AM 2 18096 3.67 3.76 3.55
12:00:01 PM 3 18094 3.90 3.56 3.48
12:10:01 PM 1 18101 2.38 3.32 3.53
12:20:01 PM 2 18113 2.78 3.01 3.21
12:30:01 PM 0 18115 1.84 2.26 2.76
12:40:01 PM 1 18134 2.30 2.41 2.61
Average: 2 17676 3.54 3.61 3.66

This number increases every day. Is there something wrong?
Is there some process increasing this plist-sz?
If yes, how to find the process is doing the plist-sz growing?

The server is new and I am affraid that Linux reachs the processes limit and stops the database.

Thanks in advance,

Neto

'ps' returns the current process table. Collected SAR data gives you cumulative statistics.
If you need to monitor the amount of processes active right now then
would tell you if there's more than a hundred active (or just 'grep ^procs_running /proc/stat'). If you need to collect such statistics remotely and without messing with SSH root logins also see SNMP (OID 1.3.6.1.2.1.25.1.6.0 IIGC) or any agents a monitoring tool like Monit, Nagios, MRTG or whatever else you favor may provide.

UnSpawn, Thanks for the answer,

But in other server the same sar -q command the number collect on column plist-sz keeps close that ones get by the ps command.

Server 2:

cat /etc/issue
Red Hat Enterprise Linux Server release 5.2 (Tikanga)

ps aux | wc -l
402

sar -q


12:40:01 PM runq-sz plist-sz ldavg-1 ldavg-5 ldavg-15
12:50:01 PM 0 486 1.18 1.46 1.06
01:00:01 PM 0 486 0.79 1.26 1.18
01:10:01 PM 0 486 1.31 1.19 1.15
01:20:01 PM 2 494 0.16 0.29 0.66
Average: 1 513 0.79 0.80 0.79

uptime
13:25:02 up 33 days, 8:25, 1 user, load average: 0.05, 0.17, 0.5

Just 100 processes more and in other server I realized the same behavior.

Thanks again,

Neto,

Compare "plist-sz" value from 'sar -q' with '\ps auxH |wc -l' wrt threads?

UnSpawn,

Thanks,

These command works, I figure out that there was a oracle processes generating a lot of threads. We reboot the 2 nodes and works.

Regards,

Neto