[Xymon] Alternate devmon SNMP

[Bruce Ferrell]

On 5/9/19 11:50 PM, W.J.M. Nelis wrote:
> Hi,
>
>> As a long time user of devmon, I'd be fascinated to see an enumeration of the bugs, features, patches etc in your repo.
>>
>> One of the biggest "bugs" I've seen, and I haven't had time to chase it down is that over time devmon seems to leak memory like a sieve, becoming quite bloated.  I run a cron 
>> job to kill (yes kill -9 ) and restart it every four hours... Not under systemd.
>
> At NLR, devmon is running on a RHEL 6.10. There is a slight memory leak, probably in xymon or devmon, which requires a restart once every 2 or 3 months. Normally, a kernel update 
> is issued by RedHat well before the memory utilisation becomes an issue.
>
> Regards,
>   Wim Nelis.

All I can tell you Wim, that's not my experience.

24 hours is enough that xymon starts sending memory use alerts and the fatcat on researching the cause IS devmon (from sourceforge) not xymon.

I'll also add that I inadvertently ran across Bruno's github repo about a year back, thinking it was a migration from sourceforge to github and not the fork that it seems to be.  I 
used it in an install... Then fell back very rapidly to the sourceforge release tarballs.  The git clone had issues running in the most basic setup. The templates... Let's just say 
I didn't have the time to figure out what all was wrong with them.  Suffice to say, that install (a clone/rebuild of an existing install) using the github repo failed in so many 
ways it was faster and easier to revert to the SF release packages.

No Bruno, I will NOT be opening issues on the github repo.  The maintainers and users on the very low volume devmon list deserve better than that and they do respond there and to 
issues on SF. Which sort of begs the question as to why this announcement was made on the xymon list and not on the devmon list.

While the SF code has some issues, that code works out of the box... To the extent SNMP works.

I need to be clear on my comment regarding SNMP;  I love the concept of SNMP and it DOES work in 98% of the cases, and it's shame it's only 98%.

BSD based systems (PfSense and OS X are the players I've worked with here... OS X even using ucd net-snmp from the Mac Ports Project) have... Interesting implementations with 
interesting quirks.  SNMP on OpenWRT is interesting too.  I've even seen Cisco do odd things.  Devmon turned ALL the quirks up and while it's fascinating sometimes to figure out 
"Why does it do THAT?",  it also points up why SNMP has never really taken solid hold in over 30 years.