perfSONAR User Q&A and Other Discussion

["Garnizov, Ivan" <>]

Hello Mike,

Indeed this tool ... actually it is a whole framework of tools (considering 
all the different services / components), is deployed mostly by research 
institutions, but is also considered and implemented (perhaps still as a 
proof of concept) by Cisco and Juniper for their high-end appliances. Anyway 
it is most useful for monitoring of end-to-end multi-domain links.

With regards the alerting and the ability to detect network issues it is the 
pS MaDDash configuration you have to review.  
http://docs.perfsonar.net/psconfig_maddash_agent.html
http://docs.perfsonar.net/maddash_config_server.html

I would like to emphasize, that the graphs behaviour and the MaDDash 
operation we discuss here in this thread are not specific to VMs.
VM operation of perfSONAR leads to a different type issues like: unreasonable 
/ unexplainable results based on the nature of VM...more specifically the 
general resource sharing and contention.

There are partners for example, which have brought pS on Docker containers 
and have a complex scheme of management, which is actually outside the scope 
of pS development.

Once you are able to understand the pS configuration and operation in 
details, it should be possible in theory to run a similar deployment like 
theirs.
Actually they are also often addressing this user list.

Regards,
Ivan Garnizov

GEANT WP6T3: pS development team
GEANT WP7T1: pS deployments GN Operations
GEANT WP9T2: Software governance in GEANT



-----Original Message-----
From:  
[] On Behalf Of Mike Cruzz
Sent: Friday, July 31, 2020 3:33 PM
To: 
Subject: Re: [perfsonar-user] Maddash WebUI does not detect VM agent down on 
dashboard when VM is powered off

Hi Ivan

Thanks very much for that. Yes I perhaps should have mentioned, this is not a 
complaint rather just feedback to see if there can be improvements made to 
PerfSonar. I really think this is a very useful tool for a network engineer 
to use and something which really delivers value more so than other 
commercial alternative products.

This is simple enough to deploy and provides valuable metrics that help to 
diagnose issues in the network.

Again my feedback is only because maybe this tool is mostly used by research 
institutions and they have vast amount of experience in being able to deploy 
and build their own custom tuning to port the data etc.

However since I am not on the developer side of the spectrum it's quite 
challenging to understand the internal for me to customize the alerting etc 
in the code.

If any enhancements can be made in future for the maddash to detect that 
would really be useful to cover a VM use case.

I'm fully aware that the recommendation to use this tool is on dedicated 
hardware however for SMB that would be cost prohibitive and an alternative VM 
solution to deploy perfSonar is a very good alternative which offers the 
necessary visibility we lack.

Thanks for the help in replying to the issue.

Regards

Mike