Kitz Forum

Broadband Related => Router Monitoring Software => Topic started by: roseway on December 05, 2012, 11:20:00 AM

Title: rs-ux and rs-w v1.0RC2
Post by: roseway on December 05, 2012, 11:20:00 AM
After a rather long interval, the issues with the previous version have been addressed and this new release candidate is available.

Changes:
- removed 2-minute overlap between successive auto snapshots
- added extra error trapping to avoid possible lockup situations
- added error log to record communication or data errors between program and router
- added detection of RSCorr values as alternative source of FEC error data
- added feature to display max/min SNRM values and times by right-clicking the SNRM graph

https://docs.google.com/folder/d/0BxbUtOYVZ_SCZ1BnY3RKYnN2b1U/edit

* My thanks to rhohne for some valuable suggestions concerning possible error situations.

(I haven't yet produced a debug version, but I will if anybody wants to try running it with GDB.)
Title: Re: rs-ux and rs-w v1.0RC2
Post by: Bald_Eagle1 on December 06, 2012, 08:13:10 AM
Hi Eric,

Quick feedback:-

It's looking very slick now.

I had it running all night, in tandem with my own 24/7 harvesting program. No clashes at all now.  :)

Just a slight "discrepancy" with FEC/RSCorr errors though? - see attached graphs, yours appears to report more than mine, but both based upon RSCorr from xdslcmd info --stats.

If anything, as my program only harvests data every 60 seconds, I would have expected higher delta values from mine, rather than from yours that harvests every 30 seconds.

Title: Re: rs-ux and rs-w v1.0RC2
Post by: roseway on December 06, 2012, 09:54:22 AM
Thanks for those comments. I think the FEC/RSCorr discrepancy is understandable - you count the number of errors in one actual minute, whereas I count the number in one sampling period and convert it to a per-minute rate. As FEC errors occur in a very bursty fashion, it's inevitable that the two methods will deliver different results.
Title: Re: rs-ux and rs-w v1.0RC2
Post by: Bald_Eagle1 on December 08, 2012, 07:46:47 AM
Quick update:-

Your program is still running continuously in tandem with mine without any clashes or glitches.
(Windows 7 Home Premium - 64 Bit)

I notice the timing of the snapshot graphs has gradually crept a little though.

The graphs were initially created at 24 minutes past the 2 hour interval.
The latest set were created at 28 minutes past.

The first set was created 5th December though & it's now 8th, so it doesn't seem like too much of an issue - until possibly the harvesting element from both programs eventually clash with each other.

Title: Re: rs-ux and rs-w v1.0RC2
Post by: roseway on December 08, 2012, 08:26:44 AM
Many thanks Paul.

4 minutes slippage in ~60 hours - I make that about 66 ms slippage per minute (33 ms per sample). I had hoped that the timer would be a bit more accurate than that, but it's rather difficult to implement a really accurate timer in a cross-platform GUI program. I'll take another look at it, but it may fall into the category of 'diminishing returns'.
Title: Re: rs-ux and rs-w v1.0RC2
Post by: broadstairs on December 10, 2012, 07:53:37 PM
Eric I'm back tonight and will give this new version a go. If I run into a hang or crash I'll let you know and then it might be an idea if I run a debug version.

Stuart
Title: Re: rs-ux and rs-w v1.0RC2
Post by: roseway on December 10, 2012, 10:35:59 PM
Thanks Stuart.
Title: Re: rs-ux and rs-w v1.0RC2
Post by: broadstairs on December 13, 2012, 11:42:14 AM
Eric last night I had a hang where I could not shutdown rsux (had to kill it) and this morning I got access violation a couple of times. I think a debug version which runs under gdb might be a good idea. (32 bit that is)

Stuart
Title: Re: rs-ux and rs-w v1.0RC2
Post by: roseway on December 13, 2012, 01:18:41 PM
OK Stuart, I'll do that later today.

One question though - did you try to shut down the program while 'Sampling' was displayed? It occurred to me after I uploaded this version that closing, stopping or pausing the program while it was in the middle of sampling was a potential problem, and I've now trapped those situations (by delaying the requested action until sampling is complete).
Title: Re: rs-ux and rs-w v1.0RC2
Post by: broadstairs on December 13, 2012, 02:10:32 PM
One question though - did you try to shut down the program while 'Sampling' was displayed? It occurred to me after I uploaded this version that closing, stopping or pausing the program while it was in the middle of sampling was a potential problem, and I've now trapped those situations (by delaying the requested action until sampling is complete).

Yes it was sampling when I clicked stop and then tried to close with the X, it said stopped but after that would not respond and had to kill it. This has happened more than once. I usually click stop before closing anyway.

Stuart
Title: Re: rs-ux and rs-w v1.0RC2
Post by: roseway on December 13, 2012, 03:36:08 PM
I've uploaded a debug version (Linux 32-bit only) to the same address:
https://docs.google.com/folder/d/0BxbUtOYVZ_SCZ1BnY3RKYnN2b1U/edit

This version includes extra error trapping to intercept pause, stop and exit commands during sampling. This should hopefully avert that particular possibility for problems.

Title: Re: rs-ux and rs-w v1.0RC2
Post by: broadstairs on December 13, 2012, 09:06:26 PM
Now running the debug version and will see what happens.

Stuart
Title: Re: rs-ux and rs-w v1.0RC2
Post by: broadstairs on December 14, 2012, 02:24:08 PM
Just a quick update to say so far I've not had any problems, I've tried stopping and closing or just closing a few times and so far no problems. However the previous version I used did not have the Stopping or Closing message come up instead of Sampling, so I think whatever changes you made in that area have had the desired effect. Now to let it run and see if I get any access violations to debug  ;)

Stuart
Title: Re: rs-ux and rs-w v1.0RC2
Post by: roseway on December 14, 2012, 03:39:41 PM
Thanks Stuart.
Title: Re: rs-ux and rs-w v1.0RC2
Post by: broadstairs on December 22, 2012, 11:35:28 AM
Been running in debug mode since last update and no problems to report at all. No hangs or access violations at all.

So now I'll run it without gdb and see what happens, as I know only too well sometimes the added code of a debug program can sometimes hide issues if you are very unlucky. However since you have had no other reports here I'm thinking it is probably good now  ;)

Stuart
Title: Re: rs-ux and rs-w v1.0RC2
Post by: roseway on December 22, 2012, 12:36:34 PM
Thanks again, Stuart.
Title: Re: rs-ux and rs-w v1.0RC2
Post by: broadstairs on December 28, 2012, 03:05:28 PM
Eric sad to say I have to report I've had two access violations today and not running gdb. The first I thought I might have caused so I ignored it - bad move... so now it's running again under gdb to see if the aforesaid problem rears its ugly head again and hopefully gives us some information on it. I thought it might have been me the first time as I had paused it and issued commands via telnet to resync and then started it again. Don't really want to force another resync just now to see if I can recreate it, but in a day or so if nothing shows up I will try.

Stuart
Title: Re: rs-ux and rs-w v1.0RC2
Post by: roseway on December 28, 2012, 04:07:53 PM
Thanks for the report, Stuart (even if it's unwelcome :) ). I'll try pausing and then forcing a resync via telnet before resuming, and see if the same happens here.
Title: Re: rs-ux and rs-w v1.0RC2
Post by: roseway on December 28, 2012, 11:13:38 PM
Stuart, it's just occurred to me that the debug version is RC3, but the non-debug version is RC2, which didn't have the extra error trapping relating to the Stop, Pause and Exit buttons. I don't want you to waste your time chasing a bug which I think is already eliminated.

[Edit] I may be talking nonsense here. I think you've been running version RC3 in both cases, first under the control of GDB, and now on its own. If this is the case, please ignore what I said above.
Title: Re: rs-ux and rs-w v1.0RC2
Post by: broadstairs on December 29, 2012, 09:50:05 AM
Stuart, it's just occurred to me that the debug version is RC3, but the non-debug version is RC2, which didn't have the extra error trapping relating to the Stop, Pause and Exit buttons. I don't want you to waste your time chasing a bug which I think is already eliminated.

[Edit] I may be talking nonsense here. I think you've been running version RC3 in both cases, first under the control of GDB, and now on its own. If this is the case, please ignore what I said above.

Eric to confirm I was running debug version but outside of gdb when I got the access violations.

Stuart
Title: Re: rs-ux and rs-w v1.0RC2
Post by: roseway on December 29, 2012, 10:21:42 AM
Yes, I was talking nonsense. My only excuse is that it was my bedtime.
Title: Re: rs-ux and rs-w v1.0RC2
Post by: broadstairs on December 29, 2012, 11:15:12 AM
Yes, I was talking nonsense. My only excuse is that it was my bedtime.
;)
Stuart
Title: Re: rs-ux and rs-w v1.0RC2
Post by: broadstairs on December 30, 2012, 01:01:45 PM
Eric I have this morning done a number of tests pausing the program while accessing the router and restarting the program including one such test where I did a reboot while paused but all worked fine, no access violations. I'll leave it running under gdb for a while longer and try some more testing in a day or so to see if I can nudge it to fail.

Stuart
Title: Re: rs-ux and rs-w v1.0RC2
Post by: roseway on December 30, 2012, 01:21:42 PM
Thanks Stuart, I appreciate all this effort you're putting in. I've also tried hard to trigger an access violation without success. I think I'm going to decide tomorrow to release the present RC3 version as a v1.0 release and spend no more time on bug hunting this version. The reason is that the code which handles all the telnet access is very untidy and much patched, making it difficult to follow through any possible coding errors. So I intend to do a complete rewrite of this part of the program to clean up the code and make it much easier to follow.

Title: Re: rs-ux and rs-w v1.0RC2
Post by: broadstairs on December 30, 2012, 01:41:29 PM
OK Eric, I'll keep it running in gdb for a few days just in case for now. You never know we might just get lucky....

Stuart