buffer remote agent results

Posted by: wolf89

buffer remote agent results - 11/14/16 11:41 AM


We use PPP to troubleshoot remote agent VPN issues.

Accordingly, we run traces from our site to their devices, but also from their location to endpoints at our site, and external to our site, via remote agent.

The problem arises when packet drops or momentary connection loss (so far always on their end) occurs.

When that happens, we get grey bars on our end - indicating loss of connectivity to the remote agent, and no useful information on the traces from our end - since all of the details disappear in the tunnel).

If the remote agent buffered even 1 to 5 minutes of trace results with timestamps, and resent that data on reconnect, it would be a huge help.

As it is, in addition to having the remote user install the remote agent, we often have to have them install the free version of pingplotter and send us snips of what is happening on their end when these events occur.
Posted by: Phillip

Re: buffer remote agent results - 11/14/16 04:27 PM

Hey wolf89,

Thanks for writing in to the forums - let me see if I can help you with this!

We are aware a bug within the remote agent, where if you are tracing to a target (or targets) and the remote agent on that device is not available (or becomes not available), and then becomes available again - it creates a bug in the trace logic and doesn't trace correctly to that target anymore.

We are definitely looking into getting this fixed for a near future point release.

In the meantime, you can get around this by clicking on the little down arrow next to the pause button, and clicking "Reset and Restart" - which allows PingPlotter to re-find the correct route again for the target.

Hopefully this helps out! If you should find yourself with any other questions, or needing any other assistance - please don't hesitate to email us at any time at support@pingplotter.com.

Have a great day!

Posted by: wolf89

Re: buffer remote agent results - 11/16/16 09:00 AM

Hi Phillip,

I wasn't really referring to the described bug (glad its being fixed tho - since you lose trace history when you have to reset and restart - I often start a separate trace and pause the crashed one so I can preserve its history)

I was referring to the limitation of the remote agent that, say in a 30 second outage between the console and remote agent:

  • The remote agent doesn't receive any instruction(s) to trace, and/or
  • the remote agent doesnt retain any data from traces it does do
  • ultimate causing no data to be displayed for the 30 second outage between the console and remote agent

I am suggesting Pingman find a way to overcome this limitation.

Obviously to accomplish this the remote agent would have to be made a little more complex and have the ability to:

  • receive and store a list of trace instructions from the console
  • act on those instructions autonomously for a certain period of time during disconnects (say 1 minute)
  • Buffer the data from those traces made during the brief 'outage'
  • transmit the buffered data back to the console (either on request or sensed reconnect)
  • and either incorporate that data back into the 'ongoing' trace at the console, or even be able to display it as a separate trace - just so that data can be viewied.

I realize all of that would add quite a bit to the remote agent, and I know its important to keep the agent as light as possible.

But the remote capability is a great tool that suffers from that one big flaw: the time you really need its data the most - during network problems between the console and remote agent (which usually means when the remote user is having some sort of issue) - its not there for you.

If its just a problem with suddenly long latencies and/or high jitter, its ok, but for a period of sustained packet loss, you're blind: was it something in the users home network? was it the user's ISP? was it coming into the edge vpn device? you dont know.

As mentioned in the original post, we've worked around the issue by having the user install a free version of pingplotter, then email snips from it back to us.

It works but its very cumbersome for everyone involved. A built-in solution would be great!

Thanks for the great tool, its been a big help to us! I'm just suggesting something that I think could make it even better.