[harlam357]

HFM.NET - Client Monitoring Application for the Folding@Home Distributed Computing Project

Homepage on Google Code: http://code.google.com/p/hfm-net/
Please see the Homepage on Google Code for the latest version.

Google Group: http://groups.google.com/group/hfm-net/
Please see the Google Group for complete release notes, help, and support.

[harlam357]

Debugging clients that appear Hung (Red Status) is easy.

Just go to the Preferences -> Defaults -> Message Level -> Change it to 'Verbose'. Then refresh one of your clients again and check the Messages Window (Tools -> Show/Hide Messages Window).

Code:

[5/30/2009-2:58:13 PM] + HFM.Instances.ClientInstance.DetermineStatus (WinSMP - HTPC)
[5/30/2009-2:58:13 PM] +  - Retrieval Time (Date) ------- : 5/30/2009 2:58:13 PM
[5/30/2009-2:58:13 PM] +  - Time Of Last Frame (TimeSpan) : 14:51:53
[5/30/2009-2:58:13 PM] +  - Offset (Minutes) ------------ : 0
[5/30/2009-2:58:13 PM] +  - Time Of Last Frame (Adjusted) : 14:51:53
[5/30/2009-2:58:13 PM] +  - Prior Day Adjustment -------- : False
[5/30/2009-2:58:13 PM] +  - Time Of Last Frame (Date) --- : 5/30/2009 2:51:53 PM
[5/30/2009-2:58:13 PM] +  - Terminal Time (Date) -------- : 5/30/2009 2:33:03 PM

You should find some output as you see above. HFM expects the machines it is monitoring to have an analogous clock (system time) setting as the machine on which HFM is running. Understandably this is not always the case. To combat this, one can setup each client with a specific 'Offset' value.

This 'Offset' is used to determine the 'Status' of the client in question. In the above example you'll see that the Time Of Last Frame is later than Terminal Time. If this is the case, the client is deemed to be running. Terminal Time is calculated differently for GPU clients vs. CPU clients. However, in either case, the Time Of Last Frame must be later than the Terminal Time. If the opposite is true, then the client will be deemed Hung and will not show any progress percentage, TPF, PPD, etc.

In the case of a notfred's VM... you will want to check the 'Client is on Virtual Machine' option. Then Debug with the times as I've stated above.

What I would recommend you do is set the system time and UTC offset on your monitored machines to the correct time. If that is done, you won't have to set any 'Offset' value for your clients. Much simpler approach.

[harlam357]

(Reserved 2)

[Adak]

Sweeeeeeeeeeeet!!

Way to go, H!

[ChasR]

It works great. The XP lag issue is a minor PITA, easily overcome by stopping the Primary GPU for a while or monitoring with a non GPU folder or a Vista machine. One of the best features that Harlam didn't mention enough is that HFM parses FahLog.txt, counts and reports the number of FINISHED, UNSTABLE_MACHINE and EARLY_UNIT_END occurrences. You can tell at a glance if you have a problem machine. THis is fantastic for monitoring the GPU2 client and even better than fantastic for monitoring a lot of clients.

[Bon3thugz43v3r]

Works great on my Vista x64 machine. Thanks again harlam for a job well done.

Question: Is my Virtual Client supposed to be red instead of green in the status? The log shows its folding but the status remains red?

[harlam357]

Is this a standard VM or a notfred Virtual Appliance?

If the latter, you need to check the box in the host config dialog that says 'Client is on Virtual Machine'. notfred VMs report the machine's local time as UTC... In other words, there is no way to specify a timezone offset in the VM. Checking that option allows HFM to compensate for that.

Otherwise, if that is a standard VM and the timezone offset is set correctly... then the time on that machine is too far behind the machine running HFM... you can fix this by either:

- Correcting the time on the Virtual Machine.

- Using the 'Client Time Offset' setting in the Host Config Dialog. If the VM's time is behind the machine running HFM by 20 minutes, set this to -20.

My method for detmining 'Status' is a bit more stringent than FahMon I think... and I personally like it that way - I believe it's more accurate (once given the right information) and does not generate the false positives that FahMon does (again, once given the right information).

[Bon3thugz43v3r]

Originally Posted by harlam357

Is this a standard VM or a notfred Virtual Appliance?

If the latter, you need to check the box in the host config dialog that says 'Client is on Virtual Machine'. notfred VMs report the machine's local time as UTC... In other words, there is no way to specify a timezone offset. Checking that option allows HFM to compensate for that.

Otherwise, if that is a standard VM and the timezone offset is set correctly... then the time on that machine is too far behind the machine running HFM... you can fix this by either:

- Correcting the time on the Virtual Machine.

- Using the 'Client Time Offset' setting in the Host Config Dialog. If the VM's time is behind the machine running HFM by 20 minutes, set this to -20.

My method for detmining 'Status' is a bit more stringent than FahMon... and I personally like it that way - I believe it's more accurate (once given the right information) and does not generate the false positives that FahMon does (again, once given the right information).

Yeah its a Notfred's VM and I was sure to check the box for virtual machine. I am not sure about the whole time offset thing.

My log in my vm reads 7:43 right now but its actually 10:43 so I guess its 3 hours behind?

I see the max value is -120 in the HFM though so I wouldn't able to fix it that way. I think I remember seeing something about getting VMtools to fix the time error?

Otherwise, awesome program I am loving it.

[Janus67]

Very well done.

[bigtoyota479]

Would you mind if I linked folks on our forum to this to try out? It looks sweet, and some of us have issues with FahMon.

[ZL1]

awesome work man

[harlam357]

Originally Posted by Bon3thugz43v3r

Yeah its a Notfred's VM and I was sure to check the box for virtual machine. I am not sure about the whole time offset thing.

My log in my vm reads 7:43 right now but its actually 10:43 so I guess its 3 hours behind?

I see the max value is -120 in the HFM though so I wouldn't able to fix it that way. I think I remember seeing something about getting VMtools to fix the time error?

Otherwise, awesome program I am loving it.

The check box for 'Client is on Virtual Machine' alone will handle the 3 hour difference due to incorrect reporting of UTC by the VM. So if you're VM log is reporting 7:43 and the machine running HFM is reading 10:43 at the same time, the 'Offset' value should be set to 0 as the only difference is the reporting of UTC... the actual minutes would then be spot on.

I set the min and max minutes to 120 respectively because if the client is that far off, then the time on that machine really needs set correctly. Really, I recommend that the time be set correctly anyway, resulting in no 'Offset' value being necessary. One will always need to check the box for 'Client is on Virtual Machine' when running a notfred VM, that cannot be helped... but the easiest thing to do with all monitored machines is just make sure they are syncing to some NTP server, then their times will be correct and you won't have any issues or need for an 'Offset' value.

Originally Posted by bigtoyota479

Would you mind if I linked folks on our forum to this to try out? It looks sweet, and some of us have issues with FahMon.

That'll be fine... I plan on putting up threads on all the Top Teams' boards. I just have to reformat my post for each forum type, and I wanted my teammates to have the first access to the app.

It's publicly available... so I can't stop you anyway. But thank you for asking.

[bigtoyota479]

If you're going to post it on everyone's forums anyway, I'll just let you do it and have the honors.

[harlam357]

Thank you kindly.

[Jolly-Swagman]

Fantastic Work there harlam357 on the HFM.NET v1.1 Biuld 10.

Sent you Email ,, but I see someone above has same issue with Notfreds VM, with the Time differences,
other than that Works like a Charm,

[klear]

All of my machines are vista 32bit and so far it works great! I really like the additional stats over FAHMon.

Thanks Harlam!

[harlam357]

Jolly, got your e-mail. What is your UTC Offset? Are you +10 or +11?

Try playing with the 'Offset' value +60 or -60 (in addition to having 'Client is on Virtual Machine' checked) for one of the clients that is showing as Hung (Red).

Some machine(s) are just off by an hour... I know, people are going to have a time with this as FahMon is less stringent on time stamps. I may have to add an option similar to 'Ignore Async Clocks' (which makes things less stringent) to make things easier on folks.

Let me know on your UTC Offset and we'll get your you straightened out.

[Jolly-Swagman]

Thanks harlam357 ,,, the UTC offset would be +10,

[jerseytomato]

Fantastic Kudos to Harlam357

Is there a way to round the numbers of the PPD and PPD/Mhz?

What do the completed #'s mean? When I started adding clients I was getting numbers from 0 to 131.

I have this running on Vista32

[ChasR]

HFM parses the log of each instance and counts the number of times the word FINISHED occurred, which represents the number of completed WUs in the current log. It also counts the number of times the words UNSTABLE and EARLY occurred, which corresponds to the number of failed WUs in the current log.