Latency / Lag Monitor on Realtime Report

[martinch]

You guys ever had issues with LAGGED agents? I have and I thought it would be a neat idea for a feature. With the feature, ViCiDial administrators will be able to monitor agents in realtime and can proactively prevent problems before they manifest in production. If an agent has a high ping, IT can be dispatched double time to resolve the laggy machine. Let me know what you guys think about that.

I've submitted the feature in it's preliminary phase on Mantis (http://www.vicidial.org/VICIDIALmantis/view.php?id=1457)which allows real-time monitoring only. If ViCiDial consider it then I would advocate for keeping some kind of history in a table on the database so that patterns can be established and resolution times can be reduced. The black hole of "what happened?" may be less of a black hole with that information to hand and can help explain issues relating to drop calls, issues with dispositioning among other things.

Thanks for considering! Discussion welcome

ViCiDial user with fast connection;



Real-time Report;



ViCiDial user with slow 3G connection;



Real-time Report;

[mflorell]

Very interesting, thanks for your contribution!

[dreedy]

awesome can't wait to see this in the svn

[martinch]

Very interesting, thanks for your contribution!

Thank you mflorell for your encouragement I did have a question about this. If this ever saw a release, do you see any use of keeping logs of agent pings and would you be open to keeping a log table for this type of thing so that ViCiDial administrators can report on this? Basically operating as another troubleshooting tool in their arsenal to get to the bottom of issues they may experience in production.

Key points to note would be;

- If we logged to the database every second, that could 21600 rows per shift per agent...which is insane which would multiply to huge numbers with more agents.
- If we logged every minute might be more sensible.
- Logging only say 1-7 days worth of ping logs before purging them either via keep alive script during the daily resets or via the db archiving script.

awesome can't wait to see this in the svn

Thanks for the feedback. I hope you find it useful if it comes out.

[jamiemurray]

I like the idea of this, I have something similar which highlights agents on a global realtime report (for all my hosted clients where we manage all aspects of their dialer), it flashes the agent row red when their screen hasn't updated within 3 seconds, I had planned to introduce live latency monitoring too, but we just pull up the SIP extension status which shows the latency for that user (not in realtime, but enough to see if latency is far too high to use the system)

As for logging, whilst I see this being a great addition for realtime, logging the data could indeed add a significant number of queries to a busy cluster, however there this could be enabled for one agent only when required or perhaps tied into the agent debugging reporting already available, just a thought really

[mflorell]

I started working on this last night and it's grown and changed quite a bit from your initial patches. Hopefully I'll have something committed soon and I'll give a more thorough description

[mflorell]

I think I got a bit carried away with this. I changed and extended it quite a bit and added more options and data collected and displayed as well as a new report. This has all been committed to svn/trunk revision 3712. Please upgrade and let me know what you think.

Take a look at the "AGENT SCREEN LATENCY" section of this new doc below for more details:

https://www.vicidial.org/docs/AGENT_SCREEN_LOGGING.txt

[martinch]

I started working on this last night and it's grown and changed quite a bit from your initial patches. Hopefully I'll have something committed soon and I'll give a more thorough description

I think I got a bit carried away with this. I changed and extended it quite a bit and added more options and data collected and displayed as well as a new report. This has all been committed to svn/trunk revision 3712. Please upgrade and let me know what you think.

Take a look at the "AGENT SCREEN LATENCY" section of this new doc below for more details:

https://www.vicidial.org/docs/AGENT_SCREEN_LOGGING.txt

It's all good, thanks Matt! I've done a code review and posted that on Mantis but there's nothing really noteworthy in my comments. I'm really glad you got time to really give this the proper treatment and flesh it out for us...I think there's a lot of potential with this feature and could really help out ViCi users across the board! Might help answer some of those unanswerable questions Thank you again.

There's quite a bit to test here but I will endeavour to get it all tested as soon as I can. Awesome work!