Page 3 of 5

Re: psummary upgrade

PostPosted: Fri Dec 18, 2009 9:29 pm
by jcoffland
toTOW wrote:Are you sure that the atoms count doesn't trigger a bug ? These WUs have more than 1M atomsn and used to show nothing in the old psummary for the atoms count because of that.

I cannot find these projects on any of the assignment servers. Do you know the IP of the WS they belong to? They may have been listed in a very stale file.

Re: psummary upgrade

PostPosted: Fri Dec 18, 2009 9:29 pm
by jcoffland
Duplicates are fixed.

Re: psummary upgrade

PostPosted: Fri Dec 18, 2009 9:30 pm
by toTOW
bollix47 wrote:
These projects have been active for months now and are currently being assigned to server 171.67.108.22. They were on psummary before you made your changes. I'm currently crunching 4 of them. They are commonly referred to as bigadv WUs.

Here's an example of one that's currently running:

Code: Select all
[14:11:24] + Attempting to get work packet
[14:11:24] - Will indicate memory of 5966 MB
[14:11:24] - Connecting to assignment server
[14:11:24] Connecting to http://assign.stanford.edu:8080/
[14:11:24] Posted data.
[14:11:24] Initial: 43AB; - Successful: assigned to (171.67.108.22).
[14:11:24] + News From Folding@Home: Welcome to Folding@Home
[14:11:24] Loaded queue successfully.
[14:11:24] Connecting to http://171.67.108.22:8080/
[14:12:03] Posted data.
[14:12:03] Initial: 0000; - Receiving payload (expected size: 30236150)
[14:12:28] - Downloaded at ~1181 kB/s
[14:12:28] - Averaged speed for that direction ~975 kB/s
[14:12:28] + Received work.
[14:12:33] Project: 2683 (Run 5, Clone 0, Gen 26)


If I remember correctly the CONTACT said kasson.

Hope this helps you track down the problem. :ewink:


bollix47 gave you the required information on the previous page ;)

Re: psummary upgrade

PostPosted: Sat Dec 19, 2009 2:57 am
by jcoffland
I found the missing projects. All the big betas were being dropped.

Re: psummary upgrade

PostPosted: Sun Dec 20, 2009 6:52 am
by Grandpa_01
All of the ATI betas are missing I believe 5724 - 5730 can't really remember the #

Re: psummary upgrade

PostPosted: Sun Dec 20, 2009 12:42 pm
by toTOW
Grandpa_01 wrote:All of the ATI betas are missing I believe 5724 - 5730 can't really remember the #


Servers 171.64.65.102 and 171.64.65.103 ;)

Re: psummary upgrade

PostPosted: Tue Dec 22, 2009 12:15 am
by lanbrown
anandhanju wrote:The GRO-PS3 projects appear to be classified as GROMACS in the new psummary page.


That was done over a month ago and not anything new.

Re: psummary upgrade

PostPosted: Wed Dec 23, 2009 10:05 pm
by harlam357
Project 2669 Core is listed as GRO-SMP. It should be GROCVS. Apologize if this has already been mentioned.

I've had to update my parsing code for HFM as well... many of the Contact fields are now empty, and they never were previously.

Thanks to Joe for getting the ProtoMol Projects listed.

Re: psummary upgrade

PostPosted: Wed Dec 23, 2009 10:39 pm
by toTOW
I think all SMP projects doesn't have the right atom count.

Also, Protomol projects are still missing (they should be listed as they are in Advmethods).

Re: psummary upgrade

PostPosted: Wed Dec 23, 2009 11:19 pm
by jcoffland
- There were more missing projects. I fixed a problem with the new psummary logic that should take care of that. Please let me know if there are any more missing projects.

Project 2669 Core is listed as GRO-SMP. It should be GROCVS. Apologize if this has already been mentioned.

- I don't know why there are so many claims that the psummary is mixing up project types. Project 2669's files clearly use the GRO-SMP project type not GROCVS. Maybe there was an error before possibly even in some third-party code.

- Regarding parsing the psummary, it should be easier now as the code is valid XML as well as valid XHTML meaning you can use an XML parser rather than an adhoc one. We will aim to keep it this way.

- The projects wo/ a contact name were introduced because I added some projects that weren't listed before in an attempt to bring back the missing projects. These projects did not list a contact name. They are gone now. However, third-party apps should allow for empty fields.

I believe psummary is back to 100%. Please let me know if it is not.

Re: psummary upgrade

PostPosted: Thu Dec 24, 2009 12:33 am
by toTOW
GRO-SMP and GROCVS were used to differentiate SMP A1 projects (GRO-SMP) and SMP A2 projects (GROCVS) ... but maybe it was only a manual hack in old psummary :?

I think that atoms count for GRO-SMP and GROCVS are still wrong ... and BigAdv projects (p268x) should also appear on psummary.

Re: psummary upgrade

PostPosted: Thu Dec 24, 2009 2:49 am
by jcoffland
toTOW wrote:GRO-SMP and GROCVS were used to differentiate SMP A1 projects (GRO-SMP) and SMP A2 projects (GROCVS) ... but maybe it was only a manual hack in old psummary :?

This was not in anything I saw.

toTOW wrote:I think that atoms count for GRO-SMP and GROCVS are still wrong

Could be. Psummary just reports what the project maintainer puts in their configuration files. That may not match what the actual simulation is doing.

toTOW wrote:and BigAdv projects (p268x) should also appear on psummary.

Those projects are there. Psummary has always only reported projects it can contact at the moment so they can go away and come back.

The psummary system is not perfect. It should be working now a little better than before. One day we will address its deficiencies more thoroughly.

Re: psummary upgrade

PostPosted: Thu Dec 24, 2009 3:55 am
by codysluder
jcoffland wrote:The psummary system is not perfect. It should be working now a little better than before. One day we will address its deficiencies more thoroughly.


Good.

Some of the 3rd party apps have addressed this shortcoming by maintaining a local history. *It really should be addressed within the psummary code. When a new project goes on-line, it should appear in the next update. When a project goes off-line, that does NOT maan that we're no longer processing the WUs. The logic should keep the project on the list for a while (ideally for the time it takes us to process the assignment, but that's much too complicated).

Re: psummary upgrade

PostPosted: Thu Dec 24, 2009 5:15 am
by harlam357
jcoffland wrote:
Project 2669 Core is listed as GRO-SMP. It should be GROCVS. Apologize if this has already been mentioned.

- I don't know why there are so many claims that the psummary is mixing up project types. Project 2669's files clearly use the GRO-SMP project type not GROCVS. Maybe there was an error before possibly even in some third-party code.

- Regarding parsing the psummary, it should be easier now as the code is valid XML as well as valid XHTML meaning you can use an XML parser rather than an adhoc one. We will aim to keep it this way.

- The projects wo/ a contact name were introduced because I added some projects that weren't listed before in an attempt to bring back the missing projects. These projects did not list a contact name. They are gone now. However, third-party apps should allow for empty fields.


p2669 == GROCVS - looks good to me :)

Had to make one slight adjustment to my code to handle any blank fields of little to no consequence on being able to monitor a project correctly - so I'm good, I'm sticking with my parser that's gotten me here, but that's good to know that it's now valid XHTML == better options for all involved and for the future.

Re: psummary upgrade

PostPosted: Thu Dec 24, 2009 8:30 am
by 7im
jcoffland wrote:
toTOW wrote:GRO-SMP and GROCVS were used to differentiate SMP A1 projects (GRO-SMP) and SMP A2 projects (GROCVS) ... but maybe it was only a manual hack in old psummary :?

This was not in anything I saw.


Not in anything you saw at Stanford? But you might remember this... FAH WIKI: Cores

For your convenience, there is a sample fahlog.txt of a GROCVS work unit here: http://foldingforum.org/viewtopic.php?p=114172#p114172 Note the core name in the queue dump at the bottom.