check processes on remote system?

Shell Programming and Scripting
1

I have a script that counts the number of oracle processes running on the system:

if [ `ps -ef | grep ora_ | wc -l` -gt 0 ]
then

and it continues based on whether or not it finds running processes. Now we would like to move oracle to a separate server, but keep the application (and this script) on the old machine.

Is there a way for a script to perform a similar check (count number of running oracle processes) on a remote host? :confused:

The systems are all HP-UX 11iv1 and shell is KSH.

TIA!! :b:

2 
if [ `ssh remote 'ps -ef | grep ora_ | wc -l'` -gt 0 ]
then

Note: grep can count!

3

you need to share ssh keys between the two servers and then you can run your command like so...

if [ `ssh server "ps -ef | grep ora_ | grep -v grep | wc -l"` -gt 0 ]
then

grep will often count and so you can remove it by grep -v grep...

4

This is a very common mistake, you don't need the second grep and wc

grep -c "[o]ra_"
5

Or:

ssh host 'pgrep oracle' && ... || ...

It seams that pgrep is not available on HP-UX, so:

ssh host 'grep [o]ra_' && ... || ...
6

When using "grep -c" to count and you use it to count how many processes are running the "grep" is a process and will get counted... if you are grep'ing to count processes then it is probably best to do "grep -v grep | wc -l" to count.

7

I think danmero meant this:

$ ps -ef|grep ora_
  oracle  3294     1  0   Sep 05 ?        1:10 ora_dbw1_xxx
  oracle 11443 11433  0 23:02:34 pts/1    0:00 grep ora_
  oracle  3292     1  0   Sep 05 ?        1:18 ora_dbw0_xxx
  oracle  3290     1  0   Sep 05 ?        0:00 ora_pmon_xxx
  oracle  3302     1  0   Sep 05 ?        0:00 ora_reco_xxx
  oracle  3304     1  0   Sep 05 ?        0:00 ora_cjq0_xxx
  oracle  3296     1  0   Sep 05 ?        2:35 ora_lgwr_xxx
  oracle  3298     1  0   Sep 05 ?        7:08 ora_ckpt_xxx
  oracle  3310     1  0   Sep 05 ?        0:02 ora_arc1_xxx
  oracle  3300     1  0   Sep 05 ?        1:32 ora_smon_xxx
  oracle  3306     1  0   Sep 05 ?        0:11 ora_qmn0_xxx
  oracle  3308     1  0   Sep 05 ?        0:05 ora_arc0_xxx
$ ps -ef|grep [o]ra_
  oracle  3294     1  0   Sep 05 ?        1:10 ora_dbw1_xxx
  oracle  3292     1  0   Sep 05 ?        1:18 ora_dbw0_xxx
  oracle  3290     1  0   Sep 05 ?        0:00 ora_pmon_xxx
  oracle  3302     1  0   Sep 05 ?        0:00 ora_reco_xxx
  oracle  3304     1  0   Sep 05 ?        0:00 ora_cjq0_xxx
  oracle  3296     1  0   Sep 05 ?        2:35 ora_lgwr_xxx
  oracle  3298     1  0   Sep 05 ?        7:08 ora_ckpt_xxx
  oracle  3310     1  0   Sep 05 ?        0:02 ora_arc1_xxx
  oracle  3300     1  0   Sep 05 ?        1:32 ora_smon_xxx
  oracle  3306     1  0   Sep 05 ?        0:11 ora_qmn0_xxx
  oracle  3308     1  0   Sep 05 ?        0:05 ora_arc0_xxx
$

Anyway, the oracle background processes will be running even if the instance is in nomount state (i.e. even if the database is not accessible).

8

sorry... i failed to acknowledge the brackets around the "o"... i never use the brackets but i might start now.. :slight_smile:

9

If you need the PID's pgrep is OK , however when you need the count grep -c do the job.
Some shell's don't like brackets without quotes(csh/tcsh).

10

Given the OP requirement pgrep (plain grep for HP-UX) should be sufficient: