Hello all, I am in a situation where I would like to stress test my servers.
From a Perl script I am trying to perform any task (preferably creating a dummy service call) a lot of times. The task must stress the database and not do something like email people lots of times which will then just measure the email server more than the Service Desk system (4.5, sp9). This must be non-interactive.
This question is related to that of Pete Reynolds the other week. I've created a script in Perl and want to know precisely what command I should use inside the loop? I've tried using sd_event but quite frankly I can't get my head around 'mapping' (the -x option). How does this work, if there is no other way of making a call from the command line. I've tried to find some documentation but can't.
Finally (sorry!), if I have two instances of a server on one machine (with 1 database shared between them), how do I send my calls to a specific instance?
There is a client command line parameter, /STRESSTEST, that might do what you want (at least, there is in 4.5). It runs all sorts of query client tasks, but I am not sure if it writes to the database at all.
If you need to use sd_event, then the .ini file is a formal description of the data that you are sending to OVSD. The mapping is what allows you to specify which fields in the .ini file map to which fields in the database.
As for deciding which instance will handle the work, I think - but am not sure - that you might be able to do something in the server config files. That is where is load balancing is set up amongst the servers. If there is load-balancing, then I think you cannot specify which server will do the work. However, if you do NOT configure with load balancing, than you just point your client to the instance of the server you want.
Hi Josiah, Thank you for your prompt reply. I'll sort your points out later. The thing is, we've tried using stress test before but the problem was that our tool to measure stress on the system (CPU usage) wasn't granular enough to see the peaks on the CPU.
I could put this as a more open ended question maybe, how can a server when it has one instance of a service installed on it be compared for speed/strain to when it has two instances (but only one database)? We're looking to see if the thread size will have an effect or not.