Kitz Forum
Broadband Related => Router Monitoring Software => Topic started by: roseway on March 19, 2013, 08:37:53 PM
-
Just a few small fixes covering the points raised in relation to v2.51.
This version was uploaded earlier today, and I'm hoping that some users will have received the automatic notification of the upgrade. If you have, can you please let me know, and if you haven't, could you shut down rs-ux/rs-w and restart it (which should trigger the notification). Thank you in advance.
Changes since v2.51:
- corrected behaviour of system tray options
- added Minimise to system tray menu
- corrected calculation of "per minute" graph values with change of sample time
- fixed crash of Windows version when the system tray options were changed before the first run of the program
- removed Linksys X3000 from the routers list (not compatible with this program)
- adjusted the layout of the tweaking page to make more room for larger fonts
- improved the "Special login" capability to make it more widely applicable (**** still experimental ****)
http://rsux.plainroad.me.uk (http://plainroad.me.uk)
[Edit: website address corrected]
-
I started up rs-ux v2.51 (64-bit), which hadn't been running for a while (by choice) and it told me that an update was available. I responded with a 'Yes, please' and the deed was done. On attempting to exit the older version, I consistently received a 'floating point exception'. :o
A new item for my wish list. Would it be possible to specify to where the new version should be downloaded, rather than assuming that the ~/.rsux/ directory would be appropriate, please?
The latest version (2.6) is now executing in a separate work-space. (And it's time for b*cat's delayed evening meal!)
-
Hello Eric,
I have good news. With your changes to the login thread, the timed option works perfectly. With the default 100ms it logs on every time. Without the timed option, it doesn't.
Attaching latest logs.
Also I'm not sure if you are aware but the http://plainroad.me.uk url in the first post does not point to the proper location.
Cheers,
Takis
EDIT: Forgot to attach the logs heh.
-
Hello Eric,
I have good news. With your changes to the login thread, the timed option works perfectly. With the default 100ms it logs on every time. Without the timed option, it doesn't.
Attaching latest logs.
Also I'm not sure if you are aware but the http://plainroad.me.uk (http://plainroad.me.uk) url in the first post does not point to the proper location.
Cheers,
Takis
EDIT: Forgot to attach the logs heh.
Thank you, that's good news.
Sorry about the incorrect URL, I've corrected it now.
-
I started up rs-ux v2.51 (64-bit), which hadn't been running for a while (by choice) and it told me that an update was available. I responded with a 'Yes, please' and the deed was done. On attempting to exit the older version, I consistently received a 'floating point exception'. :o
A new item for my wish list. Would it be possible to specify to where the new version should be downloaded, rather than assuming that the ~/.rsux/ directory would be appropriate, please?
The latest version (2.6) is now executing in a separate work-space. (And it's time for b*cat's delayed evening meal!)
Thanks for the report. I'm not sure where the floating point exception came from, but I'll look into it. Actually it may be another instance of a bug in v2.51 which has already been identified and fixed, but I will make sure of that.
This was just a first step towards (optional) auto-updating, and I'll be changing the implementation anyway. I'm going to display the changelog as part of the notification, and I can add a couple of options, such as choosing the download location, as well.
-
Thank you for considering the suggestion. :)
I suspect that the exception was a result of running rs-ux (to check for the update notification) but not starting the data-capture and graphing process because once I had received the update, I promptly attempted to exit the (earlier) binary.
-
Eric I did get the notification when I started 2.51 earlier today, however I did not let it download but went manually to find it. I would prefer it to fire up my browser and take me to the site to do the download rather than doing what you currently do.
Stuart
-
Thanks Stuart, I'll take that into account.
-
Eric,
I find it a shame that confirm close of program and exit - minimise to system tray are exclusive. I don't see why both can't be possible at once. However maybe the windows X is meant to minimise as well as the internal window red X? In that case I do understand why they need to be exclusive and no problem when it is fixed. The windows X currently just closes the program when minimise to system tray is enabled. I would have preferred the windows close as well as the red close to minimise but I don't know what is intended and maybe to keep us all happy you add yet another option for the black as well and the red close??
The time averaging seems to mostly work now but I am suspicious that I can't get a non integer value of crc/minute error even with 600 sec average. With e..g. 15 sec errors go in multiples of 4/min but they don't go in multiples of 0.25 with 240 sec or 0.1 with 600 sec. They always go in multiples of one once the average is more than 60 sec.
-
I find it a shame that confirm close of program and exit - minimise to system tray are exclusive. I don't see why both can't be possible at once.
I suppose they can both be active - I just assumed (wrongly perhaps) that this wouldn't be wanted.
However maybe the windows X is meant to minimise as well as the internal window red X? In that case I do understand why they need to be exclusive and no problem when it is fixed. The windows X currently just closes the program when minimise to system tray is enabled. I would have preferred the windows close as well as the red close to minimise but I don't know what is intended and maybe to keep us all happy you add yet another option for the black as well and the red close??
It isn't meant to happen like that, it's something I missed. My intention was that the system close button should have the same behaviour as the program close button. I'll take another look at this.
The time averaging seems to mostly work now but I am suspicious that I can't get a non integer value of crc/minute error even with 600 sec average. With e..g. 15 sec errors go in multiples of 4/min but they don't go in multiples of 0.25 with 240 sec or 0.1 with 600 sec. They always go in multiples of one once the average is more than 60 sec.
I've had a quick look at the code, and it seems that I've rounded the values unnecessarily. I'll correct this.
-
Eric I just tested the window X in Linux 32bit and sure enough it does close the program, your big red X minimises it, I'm not sure if this happened before as I've just upgraded to KDE 4.10 today so it might have changed because of that. I'm not sure if I tested that before or not.
Stuart
-
Thanks Stuart. Yes, I agree with your findings. I'd like to propose the following behaviour for the two close buttons:
The system close button
If "Confirm close of program" is enabled, then ask for confirmation before closing, otherwise close immediately.
The program exit button
If "Exit button minimises program" is enabled, then minimise immediately. In this case, if "Minimise to system tray" is also enabled, then minimise to the system tray, otherwise minimise to the task bar.
If "Exit button minimises program" is not enabled, then if "Confirm close of program" is enabled, then ask for confirmation before closing, otherwise close immediately.
I also propose to change the icon on the program exit button so that it looks different to the system close button.
(The background to this proposal is that it's actually rather difficult to make the behaviour of the system close button the same as the behaviour of the program exit button, in a way which will be consistent across all the target systems. So I'm proposing that I make them look different as well as behaving rather differently.)
-
Eric I just noticed that I have had the invalid floating point error while minimised to the system tray on v2.6 which I believe you though you had solved. Sadly I cannot get to the event log to see if anything was recorded there. I'll leave it not minimised for now as that does seem OK.
As for you post about close/minimise I think that's reasonable.
Stuart
-
Yes, I did think that one had been fixed. I'll leave it running minimised here for an extended period and see if I can get it to happen.
-
Eric, the Sagem 2504N is not supported by rs-w; it has no telnet client.
-
All running well here with the exception of intermittent errors on exit -- "unhandled exception ....", this has always happened in past versions but it does not trouble me much.
I notice that the special login offers a nice way to run commands on the router, might it be possible to have an option to only run the special login once at start up and not every time it samples? I would like to use special at the start and the normal log in during the later sampling.
Alternatively I wonder if staying logged in is a possible option i.e. only login once, and might it speed up the sample time?
I think other things like dmt stay logged in and I assume this would be OK for most people.
-
All running well here with the exception of intermittent errors on exit -- "unhandled exception ....", this has always happened in past versions but it does not trouble me much.
This really shouldn't happen, and I've been taking some time recently to investigate it more thoroughly. Hopefully there will be an improvement in the next version.
I notice that the special login offers a nice way to run commands on the router, might it be possible to have an option to only run the special login once at start up and not every time it samples? I would like to use special at the start and the normal log in during the later sampling.
That's an interesting suggestion, and it's certainly possible. I'll give it some serious thought.
Alternatively I wonder if staying logged in is a possible option i.e. only login once, and might it speed up the sample time?
I think other things like dmt stay logged in and I assume this would be OK for most people.
It would be possible, but I'm not sure that I want to do it that way because of other complications which it would introduce. It wouldn't speed up the sample time much, because the majority of the sample time is made up of the CLI commands sent after login. I'll have another think about it, but my inclination is against it.
-
Alternatively I wonder if staying logged in is a possible option i.e. only login once, and might it speed up the sample time?
I think other things like dmt stay logged in and I assume this would be OK for most people.
It would be possible, but I'm not sure that I want to do it that way because of other complications which it would introduce. It wouldn't speed up the sample time much, because the majority of the sample time is made up of the CLI commands sent after login. I'll have another think about it, but my inclination is against it.
I would caution against such an approach as it could lead to the situation of a complete 'lock up' of the modem/router, making a power-cycle the only means of recovery. :-\
-
I don't want to especially advocate staying logged in, but judging by DMT-tool which I have run for years faultlessly, it does not actually cause most people any issues.
-
I notice that the special login offers a nice way to run commands on the router, might it be possible to have an option to only run the special login once at start up and not every time it samples? I would like to use special at the start and the normal log in during the later sampling.
I've been having a play with this, and I've implemented it slightly differently by using the special login configuration on its own. If you look at the screenshot below, you'll see that response number 4 is a simple CLI command which isn't part of the login. The option "After first login..." specifies that only the first three responses will be sent to the router after the first sample in the session. This setup works with my HG612, and produces the following entries in the event log:
23 Mar 2013 23:03:22 Recording started
Connected
ATP>sh
BusyBox v1.9.1 (2010-10-15 17:59:06 CST) built-in shell (ash)
Enter 'help' for a list of built-in commands.
#
ls /bin
zebra swapdev pwd mknod ipcheck dhcps cli
xtmcmd startbsp ps mkdir ip dhcpc chown
xdslcmd sshd pppc mc igmpproxy dbase chmod
web sntp ping ls ethcmd date cat
vi sleep netstat log equipcmd cwmp busybox
upg siproxd mv ln echo cp brctl
umount sh msg klog ebtables console bftpd
tr064 rm mount kill dsldiagd cms ash
tc ripd monitor iptables dnsmasq cmfctl MidServer
#
-
Ah, a sudden thought. You should be able to use that to issue the commands to kill off Beattie's busy-body. :)
-
That's a good thought. I might make it a one-click option to add those commands.
-
The number of commands after "first" looks a neat and flexible way of arranging things. I look forward to that version being available.
-
Since v2.6 I didn't get any errors on closing anymore.
What do you think about adding a new graph to display the variation of attainable bit rates?
-
Thanks for that information.
I could certainly add that extra graph. Perhaps overlaying the data on the connection speed graph would be even more useful? I could plot it in the same colours but using dotted lines. Does anyone have a view on this?
-
Perhaps overlaying the data on the connection speed graph would be even more useful? I could plot it in the same colours but using dotted lines. Does anyone have a view on this?
Imho overlaying sounds good!
-
I want to thank you for such an excellent application, i really, really like the looks of it.
Can i please, please request one feature however (as it seems not to exist yet).
can rs-w be set to auto-start recording when the application starts? that way, i can then set it as a service and be at piece of mind knowing that as long as my server is on, its grabbing stats.
-
Is the error reporting for FEC broken or am I actually getting 15, 000+ per min? This leading to 353m in 3 days?
-
I want to thank you for such an excellent application, i really, really like the looks of it.
Can i please, please request one feature however (as it seems not to exist yet).
can rs-w be set to auto-start recording when the application starts? that way, i can then set it as a service and be at piece of mind knowing that as long as my server is on, its grabbing stats.
Thanks for your comments.
As it happens, version 2.7, which I'm about to release, recognises two command line parameters "minimise" (or "minimize") and "startrecording" which will hopefully meet your need.
-
Thanks for your comments.
As it happens, version 2.7, which I'm about to release, recognises two command line parameters "minimise" (or "minimize") and "startrecording" which will hopefully meet your need.
Spot-on! :) I only have more Q, whens it coming out!? :)
-
whens it coming out!? :)
Tomorrow. :)
-
whens it coming out!? :)
Tomorrow. :)
Awesome Sauce, that is all :)
-
Is the error reporting for FEC broken or am I actually getting 15, 000+ per min? This leading to 353m in 3 days?
As far as I know, the program accurately reflects the data reported by the router. But I'm happy to look at it again if that isn't the case. If you look at the "Telnet data" tab, under "Connection stats" you can see the raw data from the router which I use to get the data for the graphs. About half-way down there's a section headed "Total time = .....", and the FEC figures come from there. With each sample, I compare the totals shown with the corresponding values in the previous sample, and convert the difference to a "per minute" value for the graph.
-
That's where I was getting them from, so I guess I was actually getting that many. :(
Had an engineer out today getting a fair few CRC's as it's on FastPath atm, engineer due back tomorrow. How many bitswaps per min should I be seeing, is 54 healthy?
-
That's where I was getting them from, so I guess I was actually getting that many. :(
Had an engineer out today getting a fair few CRC's as it's on FastPath atm, engineer due back tomorrow. How many bitswaps per min should I be seeing, is 54 healthy?
Some routers fail to zero the total on a reboot, and have to be actually powered down to get the values zeroed.
-
That's where I was getting them from, so I guess I was actually getting that many. :(
Had an engineer out today getting a fair few CRC's as it's on FastPath atm, engineer due back tomorrow. How many bitswaps per min should I be seeing, is 54 healthy?
Some routers fail to zero the total on a reboot, and have to be actually powered down to get the values zeroed.
Thanks, but it did start at zero so I guess the 353m was true. Is 54 bitswap per min healthy or?
Cheers
-
Is 54 bitswap per min healthy or?
Based on just a few modems bitswaps/min varies a lot between each them on the same connection. On ADSL2 I have only seen values as high as 54/min with one modem and a snrm of 2.0 or less though. I think it is just errors that really matter. A lot of bitswaps shows the modem having to try hard to reduce the risk of errors
-
Thanks, but it did start at zero so I guess the 353m was true. Is 54 bitswap per min healthy or?
It's not unhealthy, but it's a fairly large figure. Together with the large numbers of FECs I would guess that either your connection is rather noisy or you're running with a low SNR margin. If the connection is stable then it's probably nothing to concern yourself about.
(For comparison, my connection is averageish, and I'm currently getting around 40 bitswaps per minute.)
[Edit] I just noticed that les-70 has already replied, and I agree with his comments.
-
54 constant throughout the day with the SNR from 6.3 to 4.4 though packet loss is my problem all my results shows it's from the outside of my gateway but hey that's BT for you. The errors have only been super high since the swap from the 80/20 to the 40/10 package.
-
I've no experience of FTTC, so I don't really know what's normal on such connections. If you want to investigate further it might be worth asking a question in the FTTC and FTTH issues section.
-
Hello Eric,
Trying to tweak SNR Margin with the slider I'm getting "Second stage login failure - check login details". I'm assuming this has something to do with me using special login.
On an unrelated note, I was trying to manually change SNR Margin via telnet and xdslctl configure --snr percentage_value does not seem to work. I can see the margin set in profile info but upon syncing SNRM is the same as before. I wonder if there is something I don't know at play here. Could this be VDSL2 related or the DSLAM not respecting the setting?
-
On an unrelated note, I was trying to manually change SNR Margin via telnet and xdslctl configure --snr percentage_value does not seem to work. I can see the margin set in profile info but upon syncing SNRM is the same as before. I wonder if there is something I don't know at play here. Could this be VDSL2 related or the DSLAM not respecting the setting?
The DSLAMs that provide the VDSL2 service for the Openreach NGA GEA will not respond to SNRM tweaking from the CPE. :(
-
[...] Could this be VDSL2 related or the DSLAM not respecting the setting?
Yes, it has also no effect on my connection because the DSLAM seems to use a fixed bitrate to synchronise.
Perhaps overlaying the data on the connection speed graph would be even more useful? I could plot it in the same colours but using dotted lines. Does anyone have a view on this?
This sounds well thought out.
-
Trying to tweak SNR Margin with the slider I'm getting "Second stage login failure - check login details". I'm assuming this has something to do with me using special login.
It looks as though there's a bug to be fixed there. I'll check it out. But in any case it appears that you won't be able to tweak your connection unfortunately.
-
I thought it would be something like that :( My ISP seems to use a fixed 40Mbps bitrate for Fastpath and 43Mbps for Interleaved which is sad because my line is rather clean..it could only get cleaner with fiber..
On ADSL2+ they were setting a 9db SNRM and I was forcing 6db pumping the bitrate to 21Mps, perfectly stable. Their profiles are always conservative following a catchall mentality which I can understand.. but locking power users out of the configuration is just sad :( Oh well!
-
I guess most ISPs aren't very keen on power users - they prefer us all to do what we're told and keep quiet.
-
I guess most ISPs aren't very keen on power users - they prefer us all to do what we're told and keep quiet.
Quoted for truth!
-
::)