[solved] monitoring Processes by name not PID

[bigturk]

Hello all, right now i have graphing signal processes, the graphs are nice the data looks good. BUT when a processes restarts for some reason or the system reboots the graph is no good, and that is because it works via PID and not process name. does anyone know of a data queries that work off process names and not PID's?
any help would be great.
thanks

[BSOD2600]

Created years ago. Searching the template repository and/or script forum will assist your quest.

http://docs.cacti.net/usertemplate:data ... ib:process

[bigturk]

thank you for your reply and i think i found your readme from another post about this same thing.
you posted in step 7.
By default, the index type will use "hrSWRunIndex, and i will need to change it to "hrSWRunName. then type in the exe name.
the only place i see hrSWRunName is in the title under Graph fields.
and where should i type in the exe. any help would be great thank you every every much.

[BSOD2600]

You don't type in the exe name anywhere. Once the data query has been added to a device, click on the 'create graphs for this host' link. Then from the list of processes, select it and click create.

[bigturk]

I'm able to get the graph up and running its when the process gets a new PID(restarts) the graph becomes useless. I have tried different combination's but no luck.
HERE is the how to:

Graph a processes CPU and memory usage on a Win32 Host.
Created by BSOD2600

1. Import all the templates.
2. copy win32_procs.xml to <cactiroot>\resource\snmp_queries\
3. Add the data query 'SNMP - Running Processes' with Re-inxed method 'Index Count Changed'. This has worked fine for me, but you might need to use 'Verify all fields'
4. Click on 'Create Graphs for this Host'. You should see a list of all the running processes on the remote computer, simular to running_procs_graph.png. Click the ones you want to graph.
5. Go to your Data Sources and open up each process you created.
6. I'd change the RRA name to something more distinct and readable (i.e. computer_proc_processname_1.rra)
7. By default, the index type will use 'hrSWRunIndex'. This number is the same as the PID in process manager. You want to change it to use 'hrSWRunName' for each process. Type in the executables name (i.e. winamp.exe). You can find the name at top of the screen for the process you're working on (i.e. computername - process - processname). It you do not change it to use the processes name, soon as your restart restart an process, its PID changes and Cacti will no longer be able to graph it! Don't forget to save the changes.
8. That should be it to make it work!

Having issues with step 7.
Thank you again for your help.

[BSOD2600]

Hmm old instructions I should update.

Basically, cacti will only reindex / update the PID of the processes if the number of them change between polling cycles. If you want to force it to do it every polling cycle regardless if its needed or not, then change the reindex method to Verify All Fields.

[bigturk]

thank you for your help and replies.
i found the area where i need to edit. under data sources, find the process graph and in the index value field type in the process name and in the index type change it to hrSWRunName and save changes. noticed that if the process goes down and gets a new PID which 99.999% of the time will,it take about 10-15 minutes before it refreshes and it continues graphing.

once again thank you for your help and support.

[sebish]

I have exact the same problem at the moment, but my server os is unix (ubuntu).

I need to monitor the services (processes) to see if specific tasks are up and running.
I now know i need to use the hrSWRunName table, but when the server restarts the oids (pids) change my graphs will crash.

Did you get it to work? And does a script for this exist for unix os and net-snmp?

[nescio]

Is this data template work in Cacti 0.8.8a as well? As I tried, It directlly gave the error (red) below

09/10/2012 10:40:32 AM - CMDPHP: Poller[0] ERROR: A DB Exec Failed!, Error:'1054', SQL:"UPDATE graph_templates_graph SET name='|host_description| - Process - |query_hrSWRunName|' WHERE local_graph_id=161'

[BSOD2600]

Yes, it does.

Does the cacti technical support page show any issues?
Any other errors in cacti.log?