combine disk volumes with one threshold for sum of disk space

[einhorn]

Hello Michael!

We met on the "Open Source Monitoring Conference" at Nürnberg.
I listened to your presentation on NsClient?++ and asked you, whether it would be possible to add a feature "One threshold for an aggregation of disk volumes" and we aggreed on a feature request.

This is what I would appreciate:
In a windows-System you have a lot of drives, sometimes with a letter, sometimes mounted somewhere in the filesystem.

I would like to check a bunch of these drives with one threshold for "free space", counting only those drives with an minimum amount of free space and which "path-String" matches one of a number of criteria.

Because I have a "complex" idea, I divide it into a core part and some extensions - they "nice to have" and less important.

Core part:

Combine a set of drives and / or volumes with one set of thresholds.
For this I need

• one or more paths to add to the filter criteria
  
• an extra threshold to exclude free spaces that are too small for use (e.g: 100 partitions with 500MB on it provide 50GB of free space. But they are useless because we need them in parts of 2GB at a minimum. Values may vary.)
  
• a set of thresholds "Warning" and "critical".
  
• Output Performance-Data: "Bytes used", perhaps also "Bytes free" (GB or other readable modifications are welcome for Volume amounts). We need these data for reporting, especially the first one.
  

Additional Part 1 - Exclusion:

If we could exclude strings from the path "search filter" we could adopt to more complex criteria (like the filter feature at the eventlog checks)

Additional Part 2 - additional performance counter:

A performance counter providing the number of partitions that match the path-criteria would be nice - just for information.

Additional Part 3 - debug output:

Normally it would be best to have short text output, especially in bigger systems. It could be nice to get a list of all matching drives / volumes in a more verbose mode.

Additional Part 4 - Provide more thresholds to minimize workload:

I would like a complex warning threshold of (10GB or 15%) free and a complex critical threshold of (5GB or 10%) free across all data volumes that fulfill the criteria above for each check.

For each warning or critical in Mode "free" if one of the conditions is met it is considered OK.
In Mode "used" it would require both to be OK.

For this I add an Excel-Sheet as an example (because the idea is more complex). Perhaps it works for you. So you can try the thresholds and values -- needed only if you would like to combine the thresholds.
--- Why this?
We work often with standard thresholds on small and big systems. If we use such a combination, we usually have less work with adapting thresholds in nagios. At this time we use it only with oracle tablespaces - there it works pretty good.

### Begin of EXAMPLE

To "paint the picture" I provide an examples (Core part plus additional part 1 and 4):

System is a database server with two database systems with SIDs "PRD" and "KON".

The data for PRD resides in drives H: and G: and in Mountpoints G:\DATA\PRD\data6 and G:\DATA\PRD\data7
Here I want to count all partitions with at least 2 GB of free space. Free Spaces that are less are assumed not to exist.

The data for KON resides in drives L: and in Mountpoints G:\DATA\KON\data6 and G:\DATA\KON\data7
Here I want to count all partitions with at least 2 GB of free space. Free Spaces that are less are assumed not to exist.

As a commandline I would appreciate something like this (I appreciate the syntax of your filters, the keywords in the example are not the same - its just an example):

CheckDriveSize_Cluster [ShowAll] MinWarnFree?=10GB MinWarnFreePercent?=15% MinCritFree?=5GB MinCritFreePercent?=10% MinBlockSize?=2GB +Drive=L: +Drive=G:\DATA\KON\data

According to Addition 1:
CheckDriveSize_Cluster [ShowAll] MinWarnFree?=10GB MinWarnFree?=15% MinCritFree?=5GB MinCritFree?=10% MinBlockSize?=2GB +Drive=H: +Drive=G: -Drive=G:\DATA\KON\

### End of EXAMPLE

I would be glad if you would add this feature to NsClient?. If I you have any questions, please feel free to ask.

Kind regards

Bernd Horn




Postscriptum:
I just saw the additional links to the wiki to explain words. If you agree to the "big picture" I could add them. I think it makes sense only after a discussion (because my wording differs a little bit from the one used in NsClient?++).

[einhorn]

Calculator for combined threshold with two values each for warning and critical - german excel version