p2675 r1 c21 g57 FAILED

[alpha754293]

Besides I don't think that it really matters or makes much of a difference.

It does make a difference because the format that 7im is suggesting is the format the mods require to check the WU database and not doing it the correct way means they have to edit it before they can do a proper search for the WU.

Here's a question: why can't we just copy&paste the information from the unitinfo.txt file and have it parse that instead?

Everything that you need ought to be in there, right?

What are the mods looking for in the WU database?

I think that if I had to cross reference it, I would just have people post their unitinfo.txt and then just read into it using like a text or character offset or something along those lines. (or however it is that you'd want to parse that file). That way, that process would be entirely automated, and since F@H already HAS the unitinfo.txt built in to each run, it ought to be a heck of a lot easier than parsing the entire log file (although, I would think that the log file ought to be used for the error messages in and of themselves).

*shrug* just a thought. (yay MATLAB?)

[Flathead74]

...Here's a question: why can't we just copy&paste the information from the unitinfo.txt file and have it parse that instead?

Everything that you need ought to be in there, right?...

In a word, no.

Troubleshooting is done using the Fahlog.txt.
The WU is identified in the Fahlog .txt in a manner that identifies the Project, Run, Clone, and Generation.
This info is not included in the unitinfo.txt.

For help in solving issues with specific WUs,
the Moderators or Administrators have access to a database that sorts WUs using the format:
Project: xxxx (Run xx, Clone xx, Gen xx)

When you simply copy and paste from the Fahlog.txt you make it easier for not only the Mods and Admins,
but also for other contributors to the project.

This forum also contains a section where you can add or search for suspected bad WUs.
If you do not use the format, Project: xxxx (Run xx, Clone xx, Gen xx), searching is much, much more difficult.
(I am pretty sure that I read complaints about search capabilities in one of your previous posts.)
Posting in the Project: xxxx (Run xx, Clone xx, Gen xx) format helps here too.

Believe it or not, some things here are tried and tested.
While it is good to remain in a state of constant improvement,
it is really not necessary to change everything, or to always go against established direction.

[alpha754293]

...Here's a question: why can't we just copy&paste the information from the unitinfo.txt file and have it parse that instead?

Everything that you need ought to be in there, right?...

In a word, no.

Troubleshooting is done using the Fahlog.txt.
The WU is identified in the Fahlog .txt in a manner that identifies the Project, Run, Clone, and Generation.
This info is not included in the unitinfo.txt.

Code: Select all

Current Work Unit
-----------------
Name: Gromacs
Tag: P2669R16C136G56
Download time: January 23 22:13:37
Due time: January 26 22:13:37
Progress: 62%  [||||||____]

See where it says "P2669R16C136G56". I beg to differ. (this is from the current WU that I'm using the benchmark the Linux systems). Interestingly enough, this nomenclature isn't GROMACS specific, but rather F@H specific. (Reference: GROMACS User Manual Version 4.0)

For help in solving issues with specific WUs,
the Moderators or Administrators have access to a database that sorts WUs using the format:
Project: xxxx (Run xx, Clone xx, Gen xx)

When you simply copy and paste from the Fahlog.txt you make it easier for not only the Mods and Admins,
but also for other contributors to the project.

This forum also contains a section where you can add or search for suspected bad WUs.
If you do not use the format, Project: xxxx (Run xx, Clone xx, Gen xx), searching is much, much more difficult.
(I am pretty sure that I read complaints about search capabilities in one of your previous posts.)
Posting in the Project: xxxx (Run xx, Clone xx, Gen xx) format helps here too.

Believe it or not, some things here are tried and tested.
While it is good to remain in a state of constant improvement,
it is really not necessary to change everything, or to always go against established direction.

Can the search accept non-alphanumeric characters in the search string? I don't think that it can.

*edit*
Native search says: project -- too many results.

Google search ignores ':'.

{p|P}2669 {r|R}16 {c|C}136 {g|G}56 ought to be synonymous with {p|P}2669{r|R}16{c|C}136{g|G}56 which ought to be synonymous with Project(:) (Run 16, Clone 136, Gen 56).

If it can't do that, then it's not a very robust system now is it?

[7im]

The TAG line is a recent addition to the unitinfo.txt file. The WU search database is much older, and at that time used the format from the fahlog.txt file.

And while yes, it would not be too difficult to have someone from Pande Group upgrade the WU search tool to accept all types of formating, it's even less work to state the preferred search format is: Project: xxxx (Run xx, Clone xx, Gen xx) and hence the preferred thread title for problematic WUs is the same, plus a few words about the problem.

Please remember that any time Pande Group spends upgrading systems that really aren't broken, is time they spend NOT working on the science of the project finding treatments and cures for diseases.

I'll quote Flathead74 here (with a slight addition of my own ), because what he said is again apropos.

Believe it or not, some things here are tried and tested. While it is good to remain in a state of constant improvement [and to question authority], it is really not necessary to change everything, or to always go against established direction.

Please also note that Mods and Admins here are volunteers. They donate what little free time they have to help you and help the project. With all the crap they have to put up with, a little courtesy (like using this preferred WU format) is the least you can do when asking for their time and help.

[alpha754293]

The TAG line is a recent addition to the unitinfo.txt file. The WU search database is much older, and at that time used the format from the fahlog.txt file.

And while yes, it would not be too difficult to have someone from Pande Group upgrade the WU search tool to accept all types of formating, it's even less work to state the preferred search format is: Project: xxxx (Run xx, Clone xx, Gen xx) and hence the preferred thread title for problematic WUs is the same, plus a few words about the problem.

Please remember that any time Pande Group spends upgrading systems that really aren't broken, is time they spend NOT working on the science of the project finding treatments and cures for diseases.

I'll quote Flathead74 here (with a slight addition of my own ), because what he said is again apropos.

Believe it or not, some things here are tried and tested. While it is good to remain in a state of constant improvement [and to question authority], it is really not necessary to change everything, or to always go against established direction.

Please also note that Mods and Admins here are volunteers. They donate what little free time they have to help you and help the project. With all the crap they have to put up with, a little courtesy (like using this preferred WU format) is the least you can do when asking for their time and help.

I suppose.

P.S. BTW...I never really got word back about what to do with the WU. As I mentioned earlier though, it got purged as my testing got underway.

[toTOW]

There's no data for it in the DB.

The interesting thing is it failed in a strange way (I don't remember seeing this before) ... it might be hardware instabilities, or a bad WU ... (we can't tell for sure until we get reports from other donors).

[alpha754293]

There's no data for it in the DB.

The interesting thing is it failed in a strange way (I don't remember seeing this before) ... it might be hardware instabilities, or a bad WU ... (we can't tell for sure until we get reports from other donors).

Really??? wow.

weird. Any chance that I can force it to pick up the WU again and re-run it to see if I can reproduce the error?

Anything official from members of the Pande Group?

[bruce]

weird. Any chance that I can force it to pick up the WU again and re-run it to see if I can reproduce the error?

No.

Even the Pande Group members have a difficult time extracting Project: xxxx (Run yy, Clone zz, Gen www) unless some special conditions are met.

After an EUE, WUs are normally reassigned (sometimes to the same client, sometimes to someone else) to validate whether it's a hardware issue or a software issue. If the next person completes it, there's a high probability you have a hardware problem (e.g.- overheating/overclocking/etc.) and if not, it is removed from circulation after a certain number of retries.

[alpha754293]

weird. Any chance that I can force it to pick up the WU again and re-run it to see if I can reproduce the error?

No.

Even the Pande Group members have a difficult time extracting Project: xxxx (Run yy, Clone zz, Gen www) unless some special conditions are met.

After an EUE, WUs are normally reassigned (sometimes to the same client, sometimes to someone else) to validate whether it's a hardware issue or a software issue. If the next person completes it, there's a high probability you have a hardware problem (e.g.- overheating/overclocking/etc.) and if not, it is removed from circulation after a certain number of retries.

It shouldn't be an overheating problem considering that it's a normal 2U rackmount system (Tyan B4882). And it doesn't support any kind of overclocking either.

Hmm...interesting.