How to fine-tune limits for filesystem tests? 🔗 link

On Mon, Feb 11, 2013 at 3:50 PM, Mills, David (IS) <user-eb64c112f0e9@xymon.invalid>wrote:

 All –

A couple of questions re: the analysis.cfg file and its syntax for the
“DISK” test:

   - The “U” suffix for warning / panic thresholds refers to **minimum**
   values for free-space available on the indicated file system, correct? In
   other words, the amount of free space has to fall below the indicated
   number of KB, right?
   - What we want is to alert when a combination of thresholds are in
   effect: the file systems are both (a) above a certain percent full, and (b)
   less than a certain amount of free space (in blocks, megs, whatever). How
   would I combine the two types of tests?


Thanks!

david

~~~~~~~~~~~~~~~~~~~
*David Mills*
Systems Administrator
*Northrop Grumman*
XXX-XXX-XXXX
user-eb64c112f0e9@xymon.invalid

From: Larry Barber [mailto:user-6ef9c2864140@xymon.invalid]
Sent: Monday, February 11, 2013 4:26 PM
To: Mills, David (IS)
Cc: xymon at xymon.com
Subject: EXT :Re: [Xymon] How to fine-tune limits for filesystem tests?

Whether a parameter is for a yellow or red alerts depends on it's position in the "DISK" line. The 'U' simply means to use the units that are reported in the test results rather the percentage used.

I don't think you can combine these like you want, and don't really understand why you would want to, after all, if you know the size of the disk you can easily convert between used percentage and amount free. You just need to set your DISK line to use the more restrictive of the two parameters.

Thanks,
Larry Barber
On Mon, Feb 11, 2013 at 3:50 PM, Mills, David (IS) <user-eb64c112f0e9@xymon.invalid<mailto:user-eb64c112f0e9@xymon.invalid>> wrote:
All -

A couple of questions re: the analysis.cfg file and its syntax for the "DISK" test:
*         The "U" suffix for warning / panic thresholds refers to *minimum* values for free-space available on the indicated file system, correct? In other words, the amount of free space has to fall below the indicated number of KB, right?
*         What we want is to alert when a combination of thresholds are in effect: the file systems are both (a) above a certain percent full, and (b) less than a certain amount of free space (in blocks, megs, whatever). How would I combine the two types of tests?

Thanks!

david

~~~~~~~~~~~~~~~~~~~
David Mills
Systems Administrator
Northrop Grumman

All -

A couple of questions re: the analysis.cfg file and its syntax for the
"DISK" test:
*       The "U" suffix for warning / panic thresholds refers to *minimum*
values for free-space available on the indicated file system, correct? In
other words, the amount of free space has to fall below the indicated
number of KB, right?
*       What we want is to alert when a combination of thresholds are in
effect: the file systems are both (a) above a certain percent full, and
(b) less than a certain amount of free space (in blocks, megs, whatever).
How would I combine the two types of tests?

Thanks!

-----Original Message-----
From: Mike Burger [mailto:user-cc5c6e80f4c5@xymon.invalid] 
Sent: Tuesday, February 12, 2013 7:32 AM
To: Mills, David (IS)
Cc: xymon at xymon.com
Subject: EXT :Re: [Xymon] How to fine-tune limits for filesystem tests?

All -

A couple of questions re: the analysis.cfg file and its syntax for the 
"DISK" test:
*       The "U" suffix for warning / panic thresholds refers to *minimum*
values for free-space available on the indicated file system, correct? 
In other words, the amount of free space has to fall below the 
indicated number of KB, right?
*       What we want is to alert when a combination of thresholds are in
effect: the file systems are both (a) above a certain percent full, 
and
(b) less than a certain amount of free space (in blocks, megs, whatever).
How would I combine the two types of tests?

Thanks!

You are correct...the U parameter is based on blocks.

If you're looking at an individual filesystem, I believe that the only way to "combine" these two parameters will be to create a custom script checks your free space and reports it back to Xymon.

Now, here's the thing...you can have certain filesystems that are reported based on actual free space (U parameter) and others, or even the default for the rest, that can be reported based on percentages.

I'm doing exactly that. My thought is that default thresholds like 90% yellow and 95% red are great for smaller filesystems...but to be quite honest, 95% full on a 500G filesystem is a far cry from being 95% full on a 5G or even a 500M filesystem.

As a result, for my larger filesystems, I choose to use a U based threshold, and then use the 90/95% thresholds for the rest.

So, if you're ok with "combining" these tests based on a filesystem by filesystem case basis, as I've done, you can do so like so:

HOST=somehost
        DISK    %^/da[0-9][0-9] 4194304U 3145728U
        DISK    /dbdump 6291456U 4194304U
        DISK    * 90 95

In the above, any filesystem that starts with /da (I have /da01 thru /da27 at present) yellow alarms under 4G free and red alarms under 3G free.
/dbdump yellow alarms under 6G and red alarms under 4G.

All other filesystems yellow alarm under 90% and red alarm under 95%.

I hope this helps.