Re: rfv(?): 5.4 ucd-snmp/memory: memTotalFree

On 27/10/06, Thomas Anders <thomas.anders@blue-cable.de> wrote:
> So it currently gives free *physical* memory for UCD-SNMP-MIB::memTotalFree
> while 5.3.x (at least on Linux where I've checked) adds free swap as the comment
> implies.

On Linux, perhaps.
But what about the other architectures.

> Is this an intended change in behaviour?

Bruce Shaw wrote:
> I don't like changing behaviours in mid-stream.


One problem with the earlier code was that the various O/S's were implemented
completely independently, and weren't altogether consistent - both in terms of
exactly what quantities particular MIB objects represented, and (for
the "cooked"
statistics) what time period averages were calculated over.

Part of the reason for moving to use the HAL modules was to consolidate things
a bit more, and try and produce a more consistent approach. Yes - this might
result in a change of behaviour for particular architectures.
Particularly where
one system was out of step with the others. But that feels (IMO) the Right
Thing To Do.

I don't mind switching to use VIRTMEM instead of PHYSMEM - but we need to
be aware that this will change behaviour for some other systems.

I vote -1 - if only as a holding mechanism.


Note that this situation can be traced back (at least in part) to the relatively
imprecise nature of the MIB descriptions.
Wes - you wrote the original MIB - what did you mean by "free memory"?
Did you mean physical or virtual memory?


Dave

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=...057&dat=121642
_______________________________________________
Net-snmp-coders mailing list
Net-snmp-coders@lists.sourceforge.net
https://lists.sourceforge.net/lists/...et-snmp-coders