Kitz Forum
Broadband Related => Router Monitoring Software => Topic started by: renluop on April 19, 2017, 03:33:24 PM
-
But having got my refurb Windows 10, I've reinstalled it, v5.9. I, however, am getting frequent invalid floating points, and often it has closed down, so my records are gappy. :(
Perhaps 5.9 is not suited to my situation, so what earlier versions would you suggest, and other advice? Please!
-
Where did you install the DSLstats files? The reason I ask is that the Windows version of DSLstats is rather fussy about its location, although I'm not entirely sure why. A couple of users have found that reinstalling DSLstats in one of the user's own folders and deleting the configuration files before restarting it, made this sort of problem go away. If you look at the event log before closing it down, you'll see a line giving the location of the snapshot folder - the configuration files are here as well, unless you've changed one or the other.
-
\User\myself\AppData\Local\dslstats\*
*files and folders contained are:
DSLStats32W5.9: files- webserver; dslstats; errors1.dat; es data; traffic; traffic; all snapshots
-
I'll have to have a think about that. It's a slightly unusual arrangement, but I can't say it's wrong.
-
I don't have Windows 10, and I can't duplicate that directory chain on Windows 8.1. But it seems an unusual arrangement. I also note that you seem to be running the webserver; do you actually need this? Its purpose is to provide a means of accessing your DSLstats graphs and text data remotely and (forgive me if I'm wrong) I doubt that you need it, and it's just an extra complication in diagnosing your problem.
I've set up DSLstats v5.9.3 from scratch on a Win 8.1 machine using normal default locations, and enabled all the "Items to Monitor" plus the webserver and uploads to MDWS, plus several snapshots. At present it's running smoothly. I would like to suggest that you do the same. While DSLstats is running, take a note of the locations of the configuration files, the snapshot files, and the webserver files (in the event log), then close DSLstats and delete all these files. Delete all the DSLstats files as well.
Now make a fresh download of v5.9 or v5.9.3 (better probably) and extract the files, including their containing directory, to C:\Program Files. Start DSLstats and set it up how you like it, but don't change any of the default folder locations.
-
That location is where I have Dslstats installed on Windows 7.
-
Thanks for that. I don't know much about Windows. What's the full pathname?
-
It just has the drive letter missing, e.g. C:
-
That directory doesn't exist on my copy of Win 8.1. It was a Dell OEM version which I bought from their surplus department, and apparently it's a bit different from the regular version.
[Edit] I've worked it out. On this system AppData is a hidden folder.
-
For the record, AppData is a hidden folder on all Windows systems. It can easily be accessed by doing Win key + R and typing 'appdata' and then hitting return. This will open the AppData folder for the currently logged in user.
-
Thanks. It's many years since I used Windows for my own computing purposes.
-
Hi Eric! You commented that my config was unusual. In view of my other troubles, might it be wise for me to adopt a more standars one? What changes do you advise I make?
-
The truth is, I know very little about Windows. As far as I know, the 'official' place to install programs is under C:\Program Files, and by default DSLstats puts its configuration files in a location which is advised by a call to the operating system. If you haven't changed it, you can see this location in the event log - it seems to vary between different versions of Windows.
When I made my suggestion, I was just trying to encourage a complete clean out of all your DSLstats files and a restart with a default setup. A couple of other people have found that doing this has resolved issues with instability on Windows 10.
Later: Please ignore this suggestion while I do some more testing
-
TBH I've lost track where I may have also posted about my problems with DSLStats! :-[
I can now say that at last I appear to have got it working properly. Four files/folders have got themselves see att) have got themselves in to AppData, the others being where suggested.
How it has worked out this time ?I don't think I did anything particularly, bar from ensuring any old odds and ends were removed.
Thanks Eric!
-
The default location for the configuration files and the snaphots is as advised by the operating system, and that's somewhere under AppData, so that's probably perfectly OK. You can see the actual locations in the first few lines of the event log.
-
Sorry but renluop should not install DSLstats in Appdata as there will be a folder created there called DSLstats and is reserved for configuration data only
-
I don't think he has installed DSLstats there. The files he refers to are also in your AppData folder.
-
Hopefully not it always works fine in C:windows 10\Users\Name\dslstats and that is as far as you should go with the install
-
Hopefully not it always works fine in C:windows 10\Users\Name\dslstats and that is as far as you should go with the install
That's where they are, bar the 4 in the attached link, that without my efforts are in my AppData\Local\dslstats. Config files are as the system placed them.
-
That's perfectly OK. The files you referred to are generated by the program when it runs, and that folder is where the OS advises configuration files should go. The file names are slightly confusing because of Windows' ridiculous default hiding of file extensions.
-
Yes that is so annoying, bit easily changed from within folder options.
https://www.howtohaven.com/system/show-file-extensions-in-windows-explorer.shtml
-
I'm just tacking this on to avoid the need for a new thread!
I had another invalid floating point error. This time I was at desk when it occurred, that was when the graphs had completed their three hour lives, and another set were setting up.
From past happenings I have had faint suspicions the error events could have happened at like moments.
A likely daft suggestion, but could my suspicions have any worth?
-
Anything is possible, but it's hard to see why. The graphs don't fill up in discrete blocks in the way you suggest - the data is stored in a continuous way, and what you see is just a scrolling section within the total stored range.
-
Hi Eric! I realised that shortly after I turned off the bedside light! My thinking was not the best even for me. :-[ :D
Maybe something with the taking of snapshots is a cause, but as you say, 'anything is possible'.
-
The best way to try to diagnose this sort of problem is to start with a very simple configuration, and slowly build up from it. I suggest that you clear out any current DSLstats files (including the configuration files) and install version 6.0.2 using the self-extracting installer. Don't change the installation folder from the location offered by the installer. Don't change anything in the configuration, apart from entering your modem login details of course. Run it like that for a couple of days if you can. After that, slowly add the features you want, testing each addition as you make it.
-
The best way to try to diagnose this sort of problem is to start with a very simple configuration, and slowly build up from it. I suggest that you clear out any current DSLstats files (including the configuration files) and install version 6.0.2 using the self-extracting installer. Don't change the installation folder from the location offered by the installer.
What if it offers the root or Program Files/ Program Files x86?
Don't change anything in the configuration, apart from entering your modem login details of course.
You mean gather no stats at all?
Run it like that for a couple of days if you can. After that, slowly add the features you want, testing each addition as you make it.
OT. P.s
I've been running from the test socket for a couple of days, but this morning had a nightmare time getting any connection, so finally I've abandoned that and connected via the hired wired extension. That produced horrific errors, SNRm fluctuations; 6, 11,-2 dB and syncs between ~7000 and ~4500 kbps.
Nothing makes sense, so I wonder if reverting to old router may provide clues.
-
What if it offers the root or Program Files/ Program Files x86?
It doesn't. It's designed to offer a suitable location.
You mean gather no stats at all?
No, by default it collects some stats, but only a selection of what's available. I suggest that you make no changes initially.
-
I've been running from the test socket for a couple of days, but this morning had a nightmare time getting any connection, so finally I've abandoned that and connected via the hired wired extension. That produced horrific errors, SNRm fluctuations; 6, 11,-2 dB and syncs between ~7000 and ~4500 kbps.
This must surely be a line fault. If you've been running from the test socket for two days and you're getting results like this, then there's little doubt. You need to report a fault to your ISP.
-
Clean installed 6.0.2 as advised, and am surprised how much is being recorded in the minimal state.
-
The default state is meant to be a useful starting point. If you can now leave it running like that for some time, that would be good.
-
will do! Also posted to ISP Forum; now wait for response!
-
As Eric recommended I have been running DSLStats 6.0.2 in the basic state, save that I added in error recording (for correspondence with ISP and of course my benefit). I have not suffered any floating point error at all.
For the ISP's fault reporting, I changed the Billion 8800NL for my reserve Tcolor 585n, which too has behaved well and stayed in sync @ 6795 kbps (Billion ~5400 kbps) with seemingly fewer errors. One oddment, probably not relevant is that the Billion reported interleaving 1/1, but the Tcolor 1/0.
I did spot some , could be, pretty regular dips in up SNRm and will look closer before posting more on here.
Edit: they must have been one-offs definitely an hour apart, 7.50 and 8.50, not seen since.
In meantime any thoughts as to what i posted are welcome.
-
The 592N has held its own, but I had some trouble getting the wireless connection to printer; the fiddling/ accidental disconnects cost me 800 kbps. That was Wed evening, i think.
I think I could have found the culprit. I had been thinking how old and untidy the doorbell equipment was; whether it should be updated. i found the input wires inserted, but one came loose in my hand.
Today the down SNR is rock steady for 75% or more of its up time, and the up SNR has varied by 1.3 dB at the most.
Does it look to you, I'm on the right track?
Now I'm musing whether to reinstall the Billion, and when. :-\
-
I think I could have found the culprit. I had been thinking how old and untidy the doorbell equipment was; whether it should be updated. i found the input wires inserted, but one came loose in my hand.
<snip>
Does it look to you, I'm on the right track?
Possibly. I am assuming the door-bell is powered through a mains-transformer? Which were the loose wires?
-
From the mains; blue lead (blue brown wiring). It's a funny beast, as the previous owner had had to strip back the outer sheath, because there is no gap in the casing for cable. He was a DIY man, so the thought. I wonder if he found it, thinking this'll do.
Price label £9.99, and it's thought to be 15-20 y.o.
-
I wouldn't waste any more time on an old and suspect doorbell system. I suggest that you disconnect it and strip out the old wiring and hardware, and replace it with a battery operated wireless doorbell. These are delightfully easy to install (i.e. no wires) and have all sorts of useful features - multiple chime units, multiple bell pushes, different tones for front door and back door, etc.
-
Eric, I've already got the replacement, a wireless one with plugthough from Screwfix. The worst bit is going to be getting the old push off the PVCu ( or whatever) doorpost. Goodness knows how it was affixed!
Still pondering on change back to Billion and when!
As to bells 'n whistles there's no need, as we're linked detached. Backdoor is only reachable via gate on wrong side, or we get in through garage and utility room.
-
I changed back to the Billion on Wednesday evening, and after some personal memory trouble with passwords got sync of 6772 kbps with 6dB SNRm. Overnight and through yesterday until about 4.30 p.m, though the noise margin was more volatile and errors greater, it seemed, than the 582N.
Around 4.30 p.m the connection dropped and was restored @ 5427 kbps. (The same sort of reduction , from memory happened with the 582N).As for some time I was away, I did not notice until late evening. From approximately 5.00 p.m. to just after 6 SNRm went beserk.
In theory I can get more by tweaking the margin, as I had done before, but am now loath to. I am using Open DNS, if relevant.
Another oddity is that although the SNR graphs have kept the same format throughout, the Y axis of the Speed graph changed. I used to be 0 -10000 kbps, but since this evening 0 – 300000 kbps. Down is almost off screen and Up has been banished to the Underworld.
Further commenti if you wish, please. Is my router probably suspect?
-
If that very suspect door-bell mains transformer is now out of the picture then I suspect there is still a problem with your circuit. I know, from what you have previously recounted, that it seems as if there are more joints than cable in your underground service feed. That could explain the "craziness" of the SNRM.
As for the change in scale of the Y-axis of the connection speed plot, I believe that there is a DSLstats option to allow it to auto-range. So if, for whatever reason, there was a large increase that would have gone "off scale" DSLstats has adjusted the display range.
-
The Conn Speed has never been above 8000, so I'd hardly expect the range to go up to 300000.
I'll see if ceasing a restarting DSLstats cures it.
-
I suggest that you just try switching that graph to logarithmic scale and then back again. That may be all you need to get it scaling correctly.
-
I think I tried that last evening. This morning I acted a i considered in my post, and that did the trick. Why that wobbly is a mystery.
-
Eric, I've already got the replacement, a wireless one with plugthough from Screwfix. The worst bit is going to be getting the old push off the PVCu ( or whatever) doorpost. Goodness knows how it was affixed!
Pull off the cover and you should find two screws holding it to the frame. :)
-
Actually I think it was stuck on somehow, but son-i-l very more hands on than I dealt with it for this old man. What takes me an hour, for him it's 5 minutes.
-
I had a look inside the extension, and it looks pretty neatly installed: Pic.
Wondering now, if I should replace (get replaced) the main and extension sockets with a filtered versions, using existing or new cable on basis it would ensure any house wiring defects would be excluded.
BTW over the weekend sync has held in @ 7043 on the 47 dB attenuation, and errors
Average error rates for 16 Jul 2017
CRC erors per hour: 8.39 Down, 5.44 Up
FEC erors per hour: 0 Down, 0 Up
HEC erors per hour: 17.9 Down, 2.72 Up
ES per hour: 9.98 Down, 4.76 Up
SES per hour: 2.50 Down, 0 Up
EditShould have not been so upbeat. Found it had come done to 5500 9.35 this morning.
-
That photo looks like a master-type socket, and having more than one master socket connected to a line probably won't help.
What master socket do you have?
-
An LJU2/1A is a master socket.
To save any more faffing about, I think we ought to see the equivalent photographs for all sockets . . . and the innards of any junction boxes.
-
That's weird! It's what has been since before we moved in 2002. Lady used to be something like the vicar's secretary, and the tidiness made me think it was all proper. The "extension" has a BT Piper local too. I think her husband had been a failed DIY man.
Our feed is d.i.g and definitely comes in to house by front door that way. ATM I can't take a pic of the real master, but it is a split one much like the stock photo from Google. If you want the innards still, I'll d/w asap.
In passing; why does having two master type sockets seem not a good idea?
Forward path? :-\
-
If you want the innards still, I'll d/w asap.
When time permits, please. Both the rear of the lower front face-plate, if there are any wires are connected to the IDCs there situated, and also the rear of main body.
In passing; why does having two master type sockets seem not a good idea?
Forward path? :-\
The circuit will have two resistive-capacitive shunts across it and potentially two over-voltage protection devices . . . But I regard it as a big indicator that there could be other non-standard wiring oddities about the circuit.
-
You mean unscrew the top half too, to see behind?
-
You mean unscrew the top half too, to see behind?
Yes please.
-
i managed to get some pics. I hope they are what's wanted. I must really encourage my fellow octogenarians to lie on the floor with screwdrivers supported by a coarse doormat. ;D
Connection has been allover place today with speed down to the 2-3000 kbps and snr shooting up to 15 for a while, and dropouts a plenty. 'suppose some my be due to my engineering skills. ::)
-
Thank you for obtaining those photographs. They help but also disclose more unknowns! :o
Looking at "mainsoc x.png" we can see that there is a connection to the IDCs on the lower front face-plate made using the "blue/white & white/blue" pair.
Looking at "mainsoc y.png" we can see that the NTE5/A is connected using the "orange and white" pair. Exactly as I would expect for either an aerial or underground service feed.
Looking ar "mainsoc z.png" we see that there appears to be three cables within the backing box. One cable contains the "orange & white" pair that feeds the NTE5/A. A second pair from that same cable appears to be "green & black"; the third pair appears to be "blue & brown". However we can also see a "red & grey" pair. To add to the confusion, there is clear evidence of more recent internal cabling present . . . CW1308 specification cable with three pairs: a "blue/white & white/blue", a "orange/white & white/orange" and a "green/white & white/green" pairs! The "blue/white & white/blue" pair is that connected to the lower front face-plate, shown in "mainsoc x.png" and noted above.
Going back to the photograph attached to Reply #44 (http://forum.kitz.co.uk/index.php/topic,19703.msg350773.html#msg350773) which shows an LJU2/1A, a master socket, in use as an extension, we see it is connected using a "brown/white & white/brown" pair. ??? I have absolutely no idea from whence that pair emanates.
It is all quite complex and confusing. But would be fun to "sort out" and fix. Unfortunately Hampshire is too far away, for this grumpy old kitteh's paws, and it is outside the trundling distance for "Walter's Wheelbarrow". :-\
I am very hesitant to say any more . . . for remote diagnosis is fraught with mishaps. Certainly I have seen enough to say that there is something abnormal with the circuit's wiring within your home. Perhaps, when time permits, Black Sheep will offer his professional opinion as to what might be done. :-\
-
The incoming feed is a standard 4 pair cable, often used for drop wires although BT seem to use cheaper single and twin pair cables these days.
By zooming in on the photograph attached to Reply #44 it appears that the socket contains lot of brown dust which is obscuring the wire colours, they all look a bit brown. It is difficult to confirm without some cleaning but I suspect the wires in use are the expected "blue/white & white/blue" pair. :)
-
Sorry, that's down to me. The original needed some adjustment, but I didn't notice the effect on one part of the pic. Here's a new one that has been dealt with selectively to give near to proper colour rendering.
Incidentally I found a site (http://www.telephonesuk.co.uk/wiring_info.htm) that might be of some interest to some
-
The incoming feed is a standard 4 pair cable, often used for drop wires although BT seem to use cheaper single and twin pair cables these days.
I think you are getting confused. For a normal residential property, an aerial drop cable will either be two or one pair. (I.e. Four or two conductors.) :-\ In this particular case, the service is via an underground feed.
By zooming in on the photograph attached to Reply #44 it appears that the socket contains lot of brown dust which is obscuring the wire colours, they all look a bit brown. It is difficult to confirm without some cleaning but I suspect the wires in use are the expected "blue/white & white/blue" pair. :)
Thank you for that helpful observation. :) As we have now subsequently seen, those "brown" wires are actually the "blue/white & white/blue" pair from a CW1308 specification cable.
-
Sorry, that's down to me. The original needed some adjustment, but I didn't notice the effect on one part of the pic. Here's a new one that has been dealt with selectively to give near to proper colour rendering.
Thank you.
With sight of the correct colours we can fairly confidentially say that the LJU2/1A is fed from the pair connected to the IDCs on the lower front face-plate. That can be conclusively confirmed by removing the lower front face-plate from the NTE5/A and testing with a telephone plugged into the LJU2/1A . . . The telephone should not work.
-
I am not confused, I admit I wasn't sure if Renluop had an underground or aerial feed but I have a couple of lengths of external drop wire type CW1417 (CAD55M) which has 4 pairs and 3 steel tension wires (donated by a helpful BT man some years ago). As I said previously it may not be used much these days on residential installations :)
-
Thank you.
With sight of the correct colours we can fairly confidentially say that the LJU2/1A is fed from the pair connected to the IDCs on the lower front face-plate. That can be conclusively confirmed by removing the lower front face-plate from the NTE5/A and testing with a telephone plugged into the LJU2/1A . . . The telephone should not work.
As conveniently soon I'll test your hypothesis, and let you know!
-
@B'cat
I've done your test. With bottom half of master removed, there's no phone on the extension.
I see in the link that I mentioned above are pics of the innards of both (https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.telephonesuk.co.uk%2Fimages%2Fsmall%2FLJU2_1A_rear_small2.jpg&hash=5defb8ce5681e99db3e695d4596ab843b9aabf10) and (https://forum.kitz.co.uk/proxy.php?request=http%3A%2F%2Fwww.telephonesuk.co.uk%2Fimages%2Fsmall%2FLJU2_rear_small2.jpg&hash=42efb7b84798cc657faf9992a26e8d6fbeb94fcb)
Were you Ren, would you risk a bit of surgery? :-\ :)
I thinking, what's he next step? Get wiring and face plates modified? Thoughts. all?
-
I've done your test. With bottom half of master removed, there's no phone on the extension.
That is good to know. So we are now in a position to assume there is no great abnormality with your own internal wiring.
What I would do, if I was assisting someone with circuit wiring like your own, would be to look in my "what-not" for either a LJU2/3A (a secondary socket) or another LJU2/1A (a master socket).
Depending upon what I had available I would either replace the existing LJU2/1A with the LJU2/3A or, with appropriate sized side-cutters, snip the two connections for each of the capacitor, the resistor and the over-voltage protection device of the existing LJU2/1A.
-
I'll have to think about that, with my two left feet on the right hand aptitude. For a start I don't know what side cutters are. Is it foolish to get someone in to put filtered face plates in, and run new cable (cat 5e) from master to extension on the chance it may improve things? Cost, but not BT? Within reason I can afford it.
-
At the moment, my understanding is that you have the modem/router situated near the extension socket and use a microfilter to make the connection. Do you have anything connected at the NTE5 (the master socket)? If yes, is that other device also connected via a microfilter?
All of the wiring from the lower front face-plate at the NTE5 to the extension socket is your own. The extension socket is also your own. (it was given to the owner of the property when the NTE5 was installed.)
To be perfectly honest, I'm not that sure if fitting a centralised filter at the NTE5 and then running data cabling from it to the vicinity of the modem/router would achieve much.
Without physically seeing and touching the current set-up, anything I suggest will just be a good guess.
Perhaps look in the classified advertisements of your local paper for an ex-GPO/ex-PO/ex-BT person who would be willing to visit, inspect and advise accordingly? :-\
-
The phone is run from the master and has a filter. Thanks for your patience, B'cat.
-
The phone is run from the master and has a filter.
That's good to know. "Looking" from afar, it is only the usage of the LJU2/1A as an extension socket that gets one of my "black looks". ;)
Thanks for your patience, B'cat.
You are welcome. :)
-
All the goings on has caused 32/4 interleaving to be applied this morning. I'm curious to know what the levels mean, even though they're no where near what you plutocrats with fibre mention time to time.
I'll wait to see if today's stability is maintained before doing anything drastic.
-
All the goings on has caused 32/4 interleaving to be applied this morning. I'm curious to know what the levels mean, . . .
I'll leave that for A.N.Other to explain as I am more "at home" with the physical infrastructure.
-
OT
I've lost all the DSLStats records over the bad period; silly me for not backing up more often!
On Saturday an a new program caused me to be locked out from logging in to Windows, so had to do a complete reset.
Where's that nine tailed whip now?