[SNMP4J] diskIOWrites - cached responses

[Frank Fock]

Hello George,

Yes, in most GETNEXT/GETBULK and GET request are instrumented
differently. Here the search seems to update the values always.
The reasoning for that behavior can be the discovery of the table
instances is not cache supported.

What seems to be strange, that a walk on diskIOWrites.5 returns
a diskIOWrites.5 instance. I would have expected an instance with OID >
diskIOWrites.5 instead.

Best regards,
Frank

George Chelidze schrieb:
> Hello Frank,
>
> Thanks for your quick reply. I have found that desired results can be 
> fetched replacing snmpget with snmpwalk:
>
> -bash-3.00$ while true; do snmpwalk -v 1 -c public localhost 
> diskIOWrites.5; sleep 1; done
> UCD-DISKIO-MIB::diskIOWrites.5 = Counter32: 755346929
> UCD-DISKIO-MIB::diskIOWrites.5 = Counter32: 755347154
> UCD-DISKIO-MIB::diskIOWrites.5 = Counter32: 755347378
> UCD-DISKIO-MIB::diskIOWrites.5 = Counter32: 755347613
> UCD-DISKIO-MIB::diskIOWrites.5 = Counter32: 755347851
> UCD-DISKIO-MIB::diskIOWrites.5 = Counter32: 755348087
> UCD-DISKIO-MIB::diskIOWrites.5 = Counter32: 755348340
> UCD-DISKIO-MIB::diskIOWrites.5 = Counter32: 755348565
> ^C
> -bash-3.00$
>
> Can it be implemented so that results are cached for GET requests and 
> not for GETNEXT requests? If I understand things correctly, snmpwalk 
> first sends GET request to the "root" of subtree, and then sends 
> GETNEXT request in turn for every subobject. From my point of view 
> there shouldn't be any difference for objects which are "leave" 
> objects in a tree (diskIOWrites.5 is a "leave", it has no subtree).
>
> Best Regards,
>
> George
>
> Frank Fock wrote:
>> Hello George,
>>
>> Neither snmpget nor the SNMP protocol or agent
>> is caching any SNMP data. Instead many implementors
>> of a specific MIB object or group of objects decide
>> to cache data for a couple of seconds (here it seems
>> to be a timeout of 5s) to avoid high load situations
>> when many clients request the same data within a
>> short interval.
>>
>> This is normal and I would not consider this a bug.
>> Rather than, it is a feature that saves some CPU
>> cycles and disk IO.
>>
>> Best regards,
>> Frank
>>
>> George Chelidze wrote:
>>> I have recently posted this question on Suns Developer Forums 
>>> "General - Maintenance" mailing list (original post: 
>>> http://forum.java.sun.com/thread.jspa?threadID=5241082&tstart=0) but 
>>> didn't received any response yet. First I thought it was sun related 
>>> problem, however after some investigation is seems to be snmp 
>>> related. That's why I decided to post here:
>>>
>>> OS: SunOS 5.10 Generic_118833-24 sun4u sparc SUNW,Sun-Fire-880
>>> NET-SNMP version:  5.0.9
>>>
>>> I am trying to monitor diskIOWrites. The problem I am faced with is 
>>> that snmpget seems to "cache" responses for the above mentioned 
>>> object. I have made the following test:
>>>
>>> -bash-3.00$ while true; do snmpget -v 1 -c public localhost 
>>> diskIOWrites.5; sleep 1; done
>>
>



-------------------------------------------------------------------------
SF.Net email is sponsored by: The Future of Linux Business White Paper
from Novell.  From the desktop to the data center, Linux is going
mainstream.  Let it simplify your IT future.
http://altfarm.mediaplex.com/ad/ck/8857-50307-18918-4
_______________________________________________
Net-snmp-users mailing list
Net-snmp-users at lists.sourceforge.net
Please see the following page to unsubscribe or change other options:
https://lists.sourceforge.net/lists/listinfo/net-snmp-users