Serverscheck consumes 1,5Gb memory
On our server, serverscheck consumes 1.5 Gb internal memory, resulting in slowing down the server because it has only 2Gb available and for 2.4 Gb processes running. After rebooting the server the situation comes back within two days. What is happening? and how can this be avoided?
This discussion has been closed.
Comments
The server has a 3.2Ghz Pentium running server 2003.
It also appears to have got rid of some of the more spurious programme errors that we were experiencing.
can you detail what processes uses what amount of memory.
Is your system a dedicated one?
we defined about 50 rules varying from checks from 60 seconds to 900 seconds.
When rebooting the server the monitoring software occupies about 0.4Gb of memory, but this seems to grow to 1.5Gb in several days.
Sincerely,
Jos
monitoring_rule thread 1 X MB
monitoring_rule thread 2 X MB
monitoring_rule thread 3 X MB
...
Could you login into the helpdesk and send your serverscheck.conf file so that we can replicate your environment.
The monitoring_manager component should not take more than around 20-40 MB of RAM instead of the almost 400 you have.
I have send the conf file to the helpdesk.
Jos
When the issue has been found / resolved then I will post an update in forum.
Jos
Can you install it on a Windows XP machine.
It is easier to say that your software does not run on a dual CPU windows 2003 server.
How can I transfer the serverscheck with its current configuration and databases to another server without typing a whole new list of checks
Jos
"It is easier to say that your software does not run on a dual CPU windows 2003 server."
=> that is a harsh statement to make. Our demo server with live screen shots runs on a Windows 2003 server.
We have done benchmarking with Intel on 4 way CPU systems and they do not reporte the same issue.
To move a configuration see knowledge base:
http://kb.serverscheck.com/index.php?page=index_v2&id=11&c=5
As we could not replicate the issue, our development team has worked hard over the weekend to include a work-around in release 6.4.4 (released this morning).
In the monitoring_watcher an additional controller is built-in that checks the memory state of the monitoring_manager.exe
If it exceeds 120 MB then it is automatically restarted. This does not affect the monitoring.