very odd - just tested it and haven't been able to reproduce the error...only thing I could suggest is to make sure you have the latest version (0.4.1.2) and delete clients.txt and dfmonsettings.txt then re-enter the settings & clients...
on dfmon i enter
CPU: test Client desc.: test Client location: \\CORE\public\distribfold
even i get this share mounted it doesnt work...
CPU: test2 Client desc.: test2 Client location: Z:\distribfold
the app settings of dfmon are untouched...
and even if i change the min. values or html output or not...
i always get that error that 37.627 is not a valid floating point value...
very odd - just tested it and haven't been able to reproduce the error...only thing I could suggest is to make sure you have the latest version (0.4.1.2) and delete clients.txt and dfmonsettings.txt then re-enter the settings & clients...
yeah its 0.4.1.2
deleting those 2 files & reentering doesnt work either...
perhaps its a problem with one of the files of the linux-client dfmon tries to read?
which files do dfmon actually read out for the values?
it reads progress.txt for the main set of values....the error you are getting means that the best energy is getting allocated to an incorrect internal field...I've tried this end to recreate the problem with my Linux box but dfMon is correctly allocating the values...not sure what is happening...
hmm interesting...
i tried now to give dfmon the full location of progress.txt
\\core\public\distribfold\progress.txt
he doesnt complain about the 37.627 but he also doesnt show up any value which is in progress.txt...it also won't update...even if i changed the interval to "check progress every 1 minute"
its just:
test test stopped 0 0.000 999999.999 (0) 0 0 0% 00:00:00 0
rofn
that would indicate it can't access the progress.txt at the path you've given (which qualifies to \\core\public\distribfold\progress.txt\progress.txt - so it is correct that dfMon can't find progress.txt)
not sure what is happening with just \\core\public\distribfold - the error you're reporting sounds like the clients.txt format is the pre 0.4.1.1 when there was 1 less entry...
try having the following as your clients.txt file:
and see what happens...Code:Test|Test|\\core\public\distribfold|1|
when i change clients.txt to
Test|Test|\\core\public\distribfold|1|
and start dfmon up it complains me about the usual error
37,627 is not a valid floating point value
when i close dfmon then and open up clients.txt again the line is changed into:
test|test|\\core\public\distribfold|1|99999999.999|0|37931,773590706|
dunno, what this 37931,773590706 value is doing in there...its not even in progress.txt...but perhaps dfmon is rounding it up in a mysterious way...i dont know...
i think progress.txt is in some wrong format for dfmon to read it out..but i don't know why...
dfmon version is 100% 0.4.1.2 (i see that in the properties of dfmon.exe in win explorer)
That last value is the time and date - I think the , is a problem (only tested with the decimal seperator as .)...looks like I need to try and work around odd decimal point seperators
Could you test something? Change the decimal seperator to . and try again - if it works then that's the issue...if not
sorry, doesn't seem to do the trick
changed the line in clients.txt to
test|test|\\core\public\distribfold|1|99999999.999|0|37931.773590706|
start dfmon
error:
37931.773590706 is not a valid floating point value
the interesting thing on this is that the line won't even show up in the gui
and i can't even close the process...i'll have to kill it via task manager...
when i try to close dfmon a error pops up with:
I/O error 32.
rofn
It is definately the decimal seperator that is causing the issue - if I change mine (under Regional Settings) to , from . then I get the 'not a valid floating point number'...time to look at changing the float to datetime functions to be multi-region
This only affects the time part of the clients.txt so shouldn't be too much work
/edit- is the best energy in progress.txt . or , seperated?
Best Energy so far: 37.627
edit: if i change my regional settings from german (austria) to english (uk) it works
seems like you got it
Last edited by rofn; 11-06-2003 at 01:43 PM.
right - the problem is that the energy value from progress.txt is always seperated by a . rather than LOCALE_SDECIMAL...I have added a tweak so that the energy value is reformatted to LOCALE_SDECIMAL if LOCALE_SDECIMAL is not .
(if that make sense )
the new version is at : http://wibble.bounceme.net/DD/DF/dfMon/locale/dfMon.exe - give that a try (make sure that the entries in clients.txt are the same as the decimal seperator in control panel - I haven't added on-the-fly regional changes yet)...
yeah this works now...
i had to change the 99999999.999 into 99999999,999 too...
dunno if this would also be updated by time into a . then...
the weird thing is that the output now looks like:
http://members.nextra.at/rofo/dfmon/output.html
isn't that output a bit wrong? dunno if dfmon does calculating of some values when it runs over a whole generation...
rofn
the output looks fine - dfMon will only monitor the best energy from gen 1 onwards....
I've changed a couple other bits (to do with the decimal seperator) so v0.4.1.3 is now out
another 'bug' squashed
yeah...0.4.1.3 works now just great...
thanks for all the help...
rofn
cool - and thanks for your help as wellOriginally posted by rofn
yeah...0.4.1.3 works now just great...
thanks for all the help...
rofn
You are right, the number that gen 0 outputs is something different not related to the RMSD/Energy. When we were doing RMSD it was fine because it would always get less than the gen 0 value so wouldn't store it. Now that we have siwtched back to just looking at the energy, the values are sometimes larger past gen 0. That is a bug I will have to fix in the next release of dfGUI. The graphs don't track gen 0 but I forgot to ignore it in the lowest energy field I guess.Originally posted by pfb
Afaik the value in gen 0 isn't used as a 'best' value by DF - it's just used to kick-start the gen 1+ process which does give a 'best' value...
Jeff.
I see you ran into the same problem that I had. Are you using Borland and AnsiString for your string arrays? Thats what got me since they automatically adjust to regional settings.Originally posted by pfb
It is definately the decimal seperator that is causing the issue - if I change mine (under Regional Settings) to , from . then I get the 'not a valid floating point number'...time to look at changing the float to datetime functions to be multi-region
Jeff.
I thought the problem sounded familiar - luckily it only affected a small bit when I was converting the energy from progress.txt to a float in order to compare it with the best so far (plus when I reset the best energy)...changing the code to check if the decimal seperator isn't '.' and change the '.' in the energy value to the right decimal seperator was fairly simple...Originally posted by Digital Parasite
I see you ran into the same problem that I had. Are you using Borland and AnsiString for your string arrays? Thats what got me since they automatically adjust to regional settings.
Jeff.
I'm using theStrToFloat Delphi function which takes into account the regional settings - shame DF doesn't use them for it's progress.txt output
---
I've also fixed the HTML output error iggy mentioned...latest version is still 0.4.1.3...
Works fine here! Thanks for the prompt bug fixes, pfb!
Minor inconvenience:
When adding the clients, DFMon doesn't write client.txt file - it waits for it's termination to do it. If the system crashes, clients have to be added again...
Any chance of saving client.txt when any change to clients description happens (including adding new client), or using save settings button to do that?
Last edited by iggy; 11-11-2003 at 08:45 AM.
iggy - I'm currently looking at that side of things as there are a few times when stats/clients might not get saved...at the moment the easiest way I have found is to write the stats/clients out every time the prog updates...
For those who are using dfMon the latest version is 0.4.1.4 - it now saves the stats every time it checks to ensure data integrity (it's been out sometime but not sure if people noticed )...can be grabbed from http://wibble.bounceme.net/DD/DF/dfMon/#Downloads
---
On another note, some work I've been doing and running CB has given me an idea for a potential enhancement for dfMon - but not sure if it would be worth coding hence me posting here for feedback on it.
Basically, the idea behind dfMon was to monitor multiple DF clients - but what about remote clients (for example, a collection of borged PCs that is located x miles away). Well, after running CB and seeing how it can report to a single monitoring app irrespective of location (plus some internet data transfer coding I've been doing), I was thinking about adding in code to dfMon which will allow a similar function. As always in cases like this, a picture (or diagram) is easier to understand so:
The idea being that each of the remote dfMon clients will transmit their stats to the local dfMon and that will report on them as 'remote' clients...this would then mean that not only would local DF clients be monitored but remote dfMon clients would as well (and those would monitor local DF clients).Code:PC PC PC PC PC PC PC PC PC PC PC PC PC PC | | | | | | | | ------ --- --------------- ------ dfMon 1 dfMon 2 dfMon 3 dfMon 4 | | | | | | | | | | | | | | | | ------------------(remote locations)-------------------- | | PC - | PC |-------------dfMon 0 PC - (local)
This would benefit those who would like to monitor their collections, intra-team gauntlets (the members would point their dfMon to a central dfMon)...and so on.
It would also mean that dfMon could run on a local PC without actually monitoring local DF clients - it would just collate the remote dfMon stats...
Obviously, there is a fair bit of work involved and so I don't want to go further with this idea if there isn't a need...so, if you like the idea let me know
---
On a similar note, I will soon be adding the extra functionality to dfMon - I hope to have v0.5 out by the end of this year
Thanks to those who are using and testing dfMon, and for the feedback from people
remote source would be excellent.
It would also be very cool if you could see points current gen, client buffered points and client total points... if your in the mood
those stats will be offered - and if the remote monitoring idea goes ahead, anything you see in dfMon would be available...Originally posted by jonnyw
remote source would be excellent.
It would also be very cool if you could see points current gen, client buffered points and client total points... if your in the mood
some of the framework for the stats is in place - I need to add some more code and test it before going further; that's why I am asking now about the remote monitoring as it will be a lot easier to add now than once the whole stats framework is in place...
for remote monitoring!
If you need any help testing it, just let me know!
yeah i'm always happy to test pre releases too. Just let me knowOriginally posted by iggy
for remote monitoring!
If you need any help testing it, just let me know!
Now been able to reproduce this - it appears when a DF client fails it causes this...(the only failure that is 100% reproducible - and causes this - is when an upload gets cut-off - this also causes dfGUI to run at 99% CPU...not sure if Jeff has seen this before/managed to trap it but now I can reproduce it I will cater for it )Originally posted by IronBits
Got an "Index out of bounds -3" popup sometime today, but it was still working.
About 24hrs and still ticking.... Not bad!
pfb, so if I get the DF client to upload and then pull the net connection part way through it will reproduce this error condition every time?
Jeff.
It seems to - that's the only way I have reproduced this problem on my setup. Still trying to figure out exactly why it is causing an issueOriginally posted by Digital Parasite
pfb, so if I get the DF client to upload and then pull the net connection part way through it will reproduce this error condition every time?
Jeff.
/edit - if you do try it, do it on a client you can afford to lose gens from as once you do it the client will be borked and a deletion of filelist.txt will be needed...
Last edited by pfb; 12-14-2003 at 01:29 PM.
why can't i edit a client? its just greyed out(edit button)
double-click the CPU text - this should populate the text boxes at the bottom with CPU, Description and the location and the Edit (and Delete) button will become active...Originally posted by N.V.M.
why can't i edit a client? its just greyed out(edit button)
If this still doesn't work let me know what OS in use and number of clients being monitored...
thanks,.... worked.
cool - still working on a better interface than the current so that things like editing are easier
few minor changes whilst I work on the bigger ones:
1) corrupt progress.txt files will now cause a status of 'Error' and the description on the HTML will say progress.txt is corrupt rather than say the meaningless error it did...
2) HTML changed...
I'm still working on the bigger changes and hope to have them implemented by April
Enjoy
http://wibble.bounceme.net/DD/DF/dfMon/#Downloads
Another update - the HTML output now supports CSS...
Details and downloads on the dfMon webpage
Looking forward to the remote monitoring!
I would really like to use dfMon, but can't quite figure it out. I'm not real bright about these things, I'll admit that.
The 2 things I don't understand is what address (or location) to put in the client-setup box in order for my "monitoring" computer to be able to see my remote computers.
Do I use the remote computers name, IP address, or what. I'm confused about this big-time.
The other thing is the entire "html" thing. Where is this file that it tells me to copy the gif's & css stuff into. Do I make the file, is it already located somewhere, or does dfmon automatically make it?
Thanks for any help.
I use the entire path to the remote directory
e.g.
\\machine_name\C$\df\distribfold
this assumes administrator access to the remote drive.
As for the html, just put all the gifs etc. in the same directory as the dfmon prog and look there for the output.html file. dfmon will create it.
Remote Monitoring Cannot wait for that
I am not a Stats Ho, it is just more satisfying to see that my numbers are better than yours.
Should be quite a few updates after Easter - currently looking at:
1) multi-threaded client reading (should improve the speed for those on a slow network)
2) remote monitoring
3) XML data storage
4) possible redesigned interface