- Monitor file sizes (got a couple of processes that dare write transaction logs that sometimes go up to 2 GB in a matter of days at which point the 2 GB filesize problem on Windows causes the application to die a horrible death)
- Monitor Directory Sizes (not so much for me but saw that pass by on the list a few days ago)
- File Age (nc_net has this) to detect hanging processes for example.
- Log parser with a read-mark (so it only reads new lines since last check). Mind you not clear on how to work this into Nagios. I mean, I would want to have nagios monitor a log file for say an error like 'ORA-xxxxx no active database connection' (or whatever the exact error is) and raise an alert at that point. What I am not sure about is how to then let Nagios know the error has been fixed. I suppose I coul parse the whole log every time and archive logs when I restart the process at which point the ORA-xxxxx would no longer be present in the log but the problem is these logs are huge and would probably constitute a huge performance hit if read completely on a regular basis. It's a tricky one.
Either way unless I am mistaken these things are not easily done in WMI if they are even possible at all. But there is no denying that WMI gives you access to a wealth of information. It will be interesting to see how you do it though. Securitywise it seems like a challenge for sure. And performance wise it is not easy either. Count me in though if you want it tested.
Cheers.
Hans
On 2/14/06, Ron Gage <[EMAIL PROTECTED]> wrote:
Quoting Ryan Wilcox <[EMAIL PROTECTED]>:
<snip>
> ron... it is understood that WMI would give users the ability to go way
> above and beyond the 'check_nt' command... what other capability are you
> interested in building in that 'check_nt' doesn't provide?
From a monitoring perspective, what more would you need than WMI? You get
direct access to the perfmon counters and hardware enumeration. You can even
enumerate services and their state via WMI (the Win32_BaseService object).
