dfMon v0.3Beta

[HaloJones]

A few suggestions:

1) At present, either the client is red (no .lock) or green (.lock) or orange (more than x mins since update)

Could red mean no .lock (client not running) and there also be a black for "cannot find client at all" (computer not accessible)?

2) Show the approximate number of points buffered

3) When dfmon fails to detect either a .lock or the client at all, record the time and display it. If the client continues to be down, leave the time unchanged. Once the client comes up, remove the time data.

[pfb]

thanks for the suggestions...

haven't had much time to get back to dfMon - my company as taken a lot of time and probably will do for the next few months...but do really want to do some work on it so might sneak a few days here and there when I can - probably be a lot of incremental releases if I do that though...

I don't use .lock at the moment but could do (in conjunction with the corrupt progress check) to do the 1st suggestion...

[HaloJones]

I started up DFMon this morning (after some muppet had killed the power to half the goddamn farm!) and it hit one of the clients and claimed "index out of bounds (3)" or somesuch. I checked that client and it seemed OK, so I deleted the client line from clients.txt and re-started DFmon. Then it did the same for another one, so I deleted that too.

Then I checked the clients directories and both had errors in the log-file coinciding with the times I'd been starting DFmon showing that the filelist.txt had been tampered with. On both clients the filelist.txt contained nothing but line-feeds - no data.

Both clients were toast and had to be re-started from scratch.

Could DFmon have had the filelist.txt files open when the DF client tried to update them resulting in the files becoming corrupt?

[pfb]

which version of dfMon was that? the latest (v 0.4.2.5) does have a check in it for a fubared client rather than just showing the Index out of bounds message...

dfMon doesn't use filelist.txt - I know dfGui does but dfMon only uses progress.txt in read-only mode for it's monitoring...sounds like when the power was cut those clients got corrupted...

[HaloJones]

You could well be right. All I know is that on the re-start, the error log showed a corrupt filelist so it could have happened when the clients were power-cycled

I'll try the latest DFmon tomorrow. Thanks for the prompt response.

[Paladin]

Howard really should make foldtrajlite update the filelist.txt file in the same way that foldit.bat gets updated -- work with a temp filelist.txt then copy it over when it's safe/complete. I and others on my team have lost significant results from corrupt filelist.txt's. Such a critical file should at least have some redundancy.

[IronBits]

It's called filelist.txt.tmp

Nothing to see here, move along...

[Paladin]

No, if it (redundancy) existed as I said there would be a filelist.bak. It's not there, so lets not "move on". It took most of us making our own upload batch files to get Howard to finally incorporate the same functionality into the main program. So if pfb adds automatic filelist.txt backup/protection to dfMon just as I have to dfManager, maybe Howard will get the idea that this is something we want as well and add it.

Therefore have been doubly on-topic since this is something for an add-on and the flow of conversation here was who was as blame for the corruption of filelist.txt -- dfMon, foldtrajlite, or the power company.

[bwkaz]

Originally posted by Paladin
No, if it (redundancy) existed as I said there would be a filelist.bak.

What does the name matter?

Howard uses filelist.txt.tmp instead of filelist.bak. It's supposed to accomplish the exact same thing that you're describing.

If it's not accomplishing that, then the fault is likely (though nobody knows for sure) your operating system having scribbled all over the file as it was (ungracefully) taken down by the power outage. Howard has no control over that at all.

[pfb]

There is a minor bug with non . decimal separators - there is a fix on the dfMon site (it only affects adding new clients)

I'm hoping to get the built-in FTP upload and points done in the next week or so (as well as a few minor bug fixes) - if anyone is still using it

[DPRGI-Federico83]

the built-in FTP is a great thing....

I have a nice idea

[pfb]

yeah - thanks for alerting me to the minor bug with the non . dp separator (must've overlooked that when I implemented the non . dp code)...

any suggestions are still welcome

[DPRGI-Federico83]

Originally posted by pfb
yeah - thanks for alerting me to the minor bug with the non . dp separator (must've overlooked that when I implemented the non . dp code)...

any suggestions are still welcome

[DPRGI-Federico83]

one minor thing...

in the output.html file could you insert a little legend of the simbols that shows the client status?

red = stopped

orange = paused

green = ok

[pfb]

there should be ALT tags for those images in the HTML output (in fact, the filenames of the 3 images are used - Running, Stalled and Stopped...)

I'll look at adding a legend as well...

[jfrye]

Just curious, can you add a linux client, or does it have to be running Samba in order to do so? Most of my folding machines are running fedora core 2....

[pfb]

dfMon will work with Samba - at the moment the executable is Windows only Haven't got a Linux box anymore to port to Kylix and test unfortunately (although should rectify that soon)...

[jfrye]

I don't want to run your client in linux, I just want to be able to add linux clients. Too bad there isn't a way to get it to work using SSH2 with a username/pass for the machine instead of setting up samba...

Anyways, thanks for the info

[Rodzilla]

remote monitoring... bring it on!!!!!!

[pfb]

currently working on improving the interface - which will make it easier to control what columns are shown, allow for sorting and also in-place editing...:



I am still testing the UI - I may release it in the next few days for wider testing but hope to include some new columns as well...

[Rodzilla]

:O you rock man! That looks slick!

[marica69]

Hi,

I have one of my machines on here that always says "stalled" with no icon. First, there is the broken link or icon missing or whatever, and second, there is the issue with the machine always being stalled. It is slow (450Mhz) but the page soes show it counting up on its generations, it is just s-l-o-w


There is always the possability of user error but I just wanted to report what I saw.

Donny

[pfb]

for the icon, have you got a Stalled.png file in the same directory as the HTML output? If not, download http://wibble.bounceme.net/DD/DF/dfMon/Stalled.png

as for the stalled issue - have you tried changing the Stalled period under the Settings menu? It is defaulted to 10 minutes but try something like 30 minutes (max is 60 minutes)...

[marica69]

The icon was not there. Apparently it did not come down with the download package. I think I pulled the whole package even though I will probably never have time to really play in the code.

I have chnaged the stall setting up to 30 as per your reccomendation, however, this machine went into stall as soon as it was installed and has remained there since ( a period of about 9 hours even thoug it has crunched and uploaded a couple of gens. (It does one about every hour or so). Will let you know what I see with the longer setting.

Thanks,
Donny

[pfb]

Originally posted by marica69
The icon was not there. Apparently it did not come down with the download package. I think I pulled the whole package even though I will probably never have time to really play in the code.

I have chnaged the stall setting up to 30 as per your reccomendation, however, this machine went into stall as soon as it was installed and has remained there since ( a period of about 9 hours even thoug it has crunched and uploaded a couple of gens. (It does one about every hour or so). Will let you know what I see with the longer setting.

Thanks,
Donny

just check the package and stalled.png wasn't in it...just changed that (must've been from when the old web server went down a few weeks back)

the other thing to check for stalled is the date/time on the PC you're monitoring and the PC dfMon is running on - what is the date/time stamp on the progress.txt file on the monitored PC?

[marica69]

The time was off on the affected machine by 32 minutes. I suspect that if I had made the stall timer longer than that it would have worked. As soon as I reset the time, the stalled notice went to working.

This software is really nice. I can now tell what is going on at a glance instead of going to each machine. Congrats on a nice piece of code.

Thanks,
Donny

[pfb]

Originally posted by marica69
The time was off on the affected machine by 32 minutes. I suspect that if I had made the stall timer longer than that it would have worked. As soon as I reset the time, the stalled notice went to working.

This software is really nice. I can now tell what is going on at a glance instead of going to each machine. Congrats on a nice piece of code.

Thanks,
Donny

glad it's sorted

[marica69]

Hi, I may have found another small nit within your program? Apparently, the time duration does not reset if the client goes from iteration zero directly to another iteration zero. This happened on one of my slow machines and the time actually appears be about a day longer than this protein has been running even though it did switch over to the new protein (showing the 98)

Not a big deal, I just always hated it when my code had this kind of thing though (but I am sortof a perfectionist when it comes to my own stuff

[pfb]

you are correct - it won't reset if it goes from gen 0 to gen 0 (at the moment)...

/edit - should be one of the things I'll be fixing, but it may never be perfect (although it will monitor the current structures done and the protein size and use that to work out if it needs to reset - the only downside is if an update doesn't change protein size)...

[pfb]

in light of the news, no more work will be done on dfMon now - but I will resurrect it when DF comes back

[Rodzilla]

Thanks for all of your hard work pfb! I look forward to seeing you again!