Bug reports for Ralph 5.52-5.54

Message boards : RALPH@home bug list : Bug reports for Ralph 5.52-5.54

To post messages, you must log in.

1 · 2 · 3 · Next

AuthorMessage
Rhiju
Volunteer moderator
Project developer
Project scientist

Send message
Joined: 14 Feb 06
Posts: 161
Credit: 3,725
RAC: 0
Message 2866 - Posted: 14 Mar 2007, 22:07:09 UTC
Last modified: 15 Mar 2007, 3:28:07 UTC

Hopefully this will fix the problem with RNA WUs running over time. Also, please post if you had difficulties with 5.51 on your machine, e.g., with graphics!
ID: 2866 · Report as offensive    Reply Quote
Profile feet1st

Send message
Joined: 7 Mar 06
Posts: 313
Credit: 116,623
RAC: 0
Message 2867 - Posted: 15 Mar 2007, 1:53:24 UTC
Last modified: 15 Mar 2007, 1:54:49 UTC

Got work? ...and what "major problem" are you addressing? I mean there have been a few quirky things going on. Just wanted to pin it down a little as to what you've found.
ID: 2867 · Report as offensive    Reply Quote
Rhiju
Volunteer moderator
Project developer
Project scientist

Send message
Joined: 14 Feb 06
Posts: 161
Credit: 3,725
RAC: 0
Message 2868 - Posted: 15 Mar 2007, 3:28:28 UTC - in response to Message 2867.  

Sorry, clarified below. Also sending out more work.
Got work? ...and what "major problem" are you addressing? I mean there have been a few quirky things going on. Just wanted to pin it down a little as to what you've found.


ID: 2868 · Report as offensive    Reply Quote
Profile feet1st

Send message
Joined: 7 Mar 06
Posts: 313
Credit: 116,623
RAC: 0
Message 2869 - Posted: 15 Mar 2007, 12:45:30 UTC

Well, the RNA tasks seemed to have the following issues:

1) Not showing models, only nstructs in the reported results (and presumably in the graphic as well?)

2) Were the steps incrementing from the start of the run through the end?

3) "stage" remaining at "initializing . . ." during entire run.

4) Producing 30 nstructs, regardless of runtime preference. Actually, in most cases this resulted in them ending "early".

5) Unusually high number of validation errors.

...and so it sounds like these 5 have all been addressed by v5.52? And I believe all of the above ONLY pertained to the new RNA work.

There have also been some problems with -107 return codes with Access Violations. Have problems in this area been addressed as well?

============= ...and the wish list...

New users are still very confused about the way percent completed and estimated runtime are presented. They end up resetting the project, and reinstalling BOINC and things. And this is the patient ones that try to get it "work properly" and post that they are having problems. We can only speculate how many others just throw up their hands and disconnect.

What are the plans for advance posting of new releases and having an advanced copy available for download? Is this only as time permits? Or a practice that is no longer required since BOINC has better compression?
ID: 2869 · Report as offensive    Reply Quote
Profile feet1st

Send message
Joined: 7 Mar 06
Posts: 313
Credit: 116,623
RAC: 0
Message 2870 - Posted: 15 Mar 2007, 13:09:28 UTC

I'm crunching model 11 of this guy:
https://ralph.bakerlab.org/result.php?resultid=457544 2f88__BOINC_INCREASE_CYCLES_RNA_ABINITIO-2f88_-_1844_55_0

And it seems to be modeling the blue end of the strand and moving it around, and the graphic shows a spec of blue popping in and out on the side opposite the blue end. Not attached to anything. No defined shape to it, just a short line. Not close enough to the rest of the action to be a sidechain.
ID: 2870 · Report as offensive    Reply Quote
Rhiju
Volunteer moderator
Project developer
Project scientist

Send message
Joined: 14 Feb 06
Posts: 161
Credit: 3,725
RAC: 0
Message 2871 - Posted: 15 Mar 2007, 18:37:04 UTC - in response to Message 2869.  

Hi feet1st, thanks for summarizing the issues. I've tried to put in fixes for all five RNA issues. I did put in a fix for a known problem withthe graphics (where it gets confused about drawing the last segment of the molecule), not sure if that will solve the -107 issue.

Let me know if that is helping. Certainly, the validate rate seems higher from my end! I'll send out more work, too.

As for advance copies for download on r@h, we've now compressed the app for Windows and Linux, so that seems to address the bandwidth issue for (most) users with multiple machines. I think we had trouble compressing the Mac builds, but if you point us to a good compressor, we'll try it of course!


Well, the RNA tasks seemed to have the following issues:

1) Not showing models, only nstructs in the reported results (and presumably in the graphic as well?)

2) Were the steps incrementing from the start of the run through the end?

3) "stage" remaining at "initializing . . ." during entire run.

4) Producing 30 nstructs, regardless of runtime preference. Actually, in most cases this resulted in them ending "early".

5) Unusually high number of validation errors.

...and so it sounds like these 5 have all been addressed by v5.52? And I believe all of the above ONLY pertained to the new RNA work.

There have also been some problems with -107 return codes with Access Violations. Have problems in this area been addressed as well?

============= ...and the wish list...

New users are still very confused about the way percent completed and estimated runtime are presented. They end up resetting the project, and reinstalling BOINC and things. And this is the patient ones that try to get it "work properly" and post that they are having problems. We can only speculate how many others just throw up their hands and disconnect.

What are the plans for advance posting of new releases and having an advanced copy available for download? Is this only as time permits? Or a practice that is no longer required since BOINC has better compression?


ID: 2871 · Report as offensive    Reply Quote
Rhiju
Volunteer moderator
Project developer
Project scientist

Send message
Joined: 14 Feb 06
Posts: 161
Credit: 3,725
RAC: 0
Message 2872 - Posted: 15 Mar 2007, 21:37:41 UTC - in response to Message 2871.  

I'm seeing a freaky blue dot showing up at random spots on 1qxa_RNA workunits. Let me see if I can find the fix today. Any other problems showing up out there?
ID: 2872 · Report as offensive    Reply Quote
Profile feet1st

Send message
Joined: 7 Mar 06
Posts: 313
Credit: 116,623
RAC: 0
Message 2873 - Posted: 15 Mar 2007, 22:34:38 UTC - in response to Message 2872.  

Great! Glad to hear they are ALL addressed. I realize several of them were not "major" problems, but they each caused a number of concerned reports on the boards from confused observers.


I'm seeing a freaky blue dot showing up at random spots on 1qxa_RNA workunits. Let me see if I can find the fix today. Any other problems showing up out there?


...and 1kka__...RNA...
...and 2f88__...RNA... (I've got a second 2f88 and am seeing the same "freaky blue dot" on it as well).
ID: 2873 · Report as offensive    Reply Quote
Profile feet1st

Send message
Joined: 7 Mar 06
Posts: 313
Credit: 116,623
RAC: 0
Message 2874 - Posted: 16 Mar 2007, 13:16:58 UTC

Since noone seems to have any other problems...

Wondering how difficult it might be to change the preferences page so that when a runtime preference is established, that a "maximum runtime allowed" field is shown. And have it just show 5x the amount of the preference (I believe 5x is the point at which the watchdog would end a task).

I KNOW the screen says "target", but many people seem to consider that an absolute "limit". By showing a maximum as well, it would make very clear what the project considers the limit to be.

Otherwise, how difficult would it be to predict that a given machine will not complete a model for a given task prior to it's configured preferred runtime? If there were some way to avoid sending a long running DOC task to a slow PC with a short preference, that would be another way to minimize the problem. But then you've got other users that do understand how it works, and just want to get results back to you faster and make good use of their old slow machine. If the machine were slow enough, it would be entirely possible that no tasks would complete within the preferred RT.
ID: 2874 · Report as offensive    Reply Quote
Profile Conan
Avatar

Send message
Joined: 16 Feb 06
Posts: 364
Credit: 1,368,421
RAC: 0
Message 2875 - Posted: 17 Mar 2007, 0:48:37 UTC

Have 6 workunits that have sat on 2 of my computers for many, many hours but have not progressed.
I had to abort all of them. 4 were on the one machine which had all 4 cores alocated to WU's but the time was not moving the cpus were on Zero usage, the other computer had 2 WU's going but they also showed no cpu usage.
On Boinc Manager it showed over an hour (17% -19%) done for two of the units and over 4 hours (66%-69%) done on the other 4 units but in my results it shows Zero cpu time, how can this be?

https://ralph.bakerlab.org/result.php?resultid=457366 (17% complete)
https://ralph.bakerlab.org/result.php?resultid=457365 (17%+ complete)

https://ralph.bakerlab.org/result.php?resultid=458259 (66%+ complete)
https://ralph.bakerlab.org/result.php?resultid=458261 (66%+ complete)
https://ralph.bakerlab.org/result.php?resultid=458336 (66%+ complete)
https://ralph.bakerlab.org/result.php?resultid=458337 (66%+ complete)


ID: 2875 · Report as offensive    Reply Quote
Thomas Leibold

Send message
Joined: 25 Feb 07
Posts: 27
Credit: 77,464
RAC: 0
Message 2876 - Posted: 17 Mar 2007, 3:33:11 UTC
Last modified: 17 Mar 2007, 3:33:36 UTC

I thought the linux client was statically linked ? I'm getting an error because of a missing shared library (C++ Standard Library version 6) with the 5.52 Ralph client.

<stderr_txt>
rosetta_beta_5.52_i686-pc-linux-gnu: error while loading shared libraries: libstdc++.so.6: cannot open shared object file: No such file or directory

</stderr_txt>

Workunits 408735, 408642, 408607
OS: SuSE Linux 9.3
ID: 2876 · Report as offensive    Reply Quote
Ty

Send message
Joined: 18 Feb 06
Posts: 1
Credit: 4,431
RAC: 0
Message 2877 - Posted: 17 Mar 2007, 5:44:08 UTC

Work Unit 408043 rosetta_beta 5.52 Boinc 5.4.11
Before starting work, computer showed starting time to completion 54.xx min.
While computer worked on work unit, time to completion incremented, instead of decrementing and progress remained at 1.000% until completion.
Computer finished work unit in 2368.02 seconds, about 39.5 minutes,
and displayed progress at 100% and uploaded.
Computer: AMD Sempron 3400+ Win XP Pro x64 SP2
ID: 2877 · Report as offensive    Reply Quote
Conrad Poohs
Avatar

Send message
Joined: 29 Aug 06
Posts: 9
Credit: 1,955
RAC: 0
Message 2878 - Posted: 17 Mar 2007, 10:23:20 UTC

This WU ran for two hours, my setting is four, and produced 30 nstructs 1 decoys.
ID: 2878 · Report as offensive    Reply Quote
LudwigVonDrake

Send message
Joined: 9 Aug 06
Posts: 1
Credit: 46,669
RAC: 0
Message 2879 - Posted: 17 Mar 2007, 15:36:41 UTC

I'm getting this error also.

"I thought the linux client was statically linked ? I'm getting an error because of a missing shared library (C++ Standard Library version 6) with the 5.52 Ralph client.

<stderr_txt>
rosetta_beta_5.52_i686-pc-linux-gnu: error while loading shared libraries: libstdc++.so.6: cannot open shared object file: No such file or directory

</stderr_txt>
"
ID: 2879 · Report as offensive    Reply Quote
John Hunt
Avatar

Send message
Joined: 16 Mar 07
Posts: 10
Credit: 28,654
RAC: 0
Message 2880 - Posted: 17 Mar 2007, 17:24:19 UTC
Last modified: 17 Mar 2007, 18:19:02 UTC

This WU ran for almost two hours (although WU length in my profile was at default = 1 hr) showing 1% complete all the way until it completed successfully.

<edit>
Looks like the same thing is going to happen with this WU. I'll let it run.




ID: 2880 · Report as offensive    Reply Quote
John Hunt
Avatar

Send message
Joined: 16 Mar 07
Posts: 10
Credit: 28,654
RAC: 0
Message 2881 - Posted: 17 Mar 2007, 19:16:45 UTC


<edit>
Looks like the same thing is going to happen with this WU. I'll let it run.


WU finished in 51 mins - % complete was at 1% all the way through.


ID: 2881 · Report as offensive    Reply Quote
Profile [B^S] JoeB@Ky

Send message
Joined: 11 Oct 06
Posts: 8
Credit: 39,098
RAC: 0
Message 2882 - Posted: 17 Mar 2007, 21:03:01 UTC

Every WU I have run under 5.52 has stuck at 1.000% completed no matter how long I have let them run. After 2-3 hours i have been abourting them. No problems with any of the previous sets of WU's.
ID: 2882 · Report as offensive    Reply Quote
Profile anders n

Send message
Joined: 16 Feb 06
Posts: 166
Credit: 131,419
RAC: 0
Message 2884 - Posted: 18 Mar 2007, 8:16:34 UTC
Last modified: 18 Mar 2007, 9:15:48 UTC

Bug ?

I have one WU https://ralph.bakerlab.org/result.php?resultid=463216
that now is at 7H 15min on a pref. of 4 H.
It is at 76.9% and has been on that for the last 2H at least.
I can't get it to show grafics.
I have 2 cores on that MAC and on the other core there is a Rosetta running
where I can se grafics ok.

I'll let it run and se how it turns out.

Anders n

[edit] 8H 15 min same 76.9% [/edit]
ID: 2884 · Report as offensive    Reply Quote
Thomas Leibold

Send message
Joined: 25 Feb 07
Posts: 27
Credit: 77,464
RAC: 0
Message 2885 - Posted: 18 Mar 2007, 19:20:30 UTC - in response to Message 2876.  

<stderr_txt>
rosetta_beta_5.52_i686-pc-linux-gnu: error while loading shared libraries: libstdc++.so.6: cannot open shared object file: No such file or directory
</stderr_txt>


Looks like my computer tried to get some Ralph work done last night: at least 15 of those errors.
I have installed libstdc++.so.6.0.3 (there is no official package for SuSE 9.3, but I found a third party package that happened to include this library because they needed it too).
Of course now the project is out of work, so I don't know whether that would have solved the problem or if there are other shared libraries that are missing as well.
ID: 2885 · Report as offensive    Reply Quote
Michael.L

Send message
Joined: 26 Nov 06
Posts: 5
Credit: 1,173
RAC: 0
Message 2886 - Posted: 18 Mar 2007, 22:49:59 UTC
Last modified: 18 Mar 2007, 22:59:20 UTC

18/03/2007 22:39:55|ralph@home|Starting BENCH_04JUMPING_SAVE_ALL_OUT_-1hz6A-_NATIVE_PAIR_7_57_BARCODE_R51H_1845_25_1
18/03/2007 22:39:56|ralph@home|Starting task BENCH_04JUMPING_SAVE_ALL_OUT_-1hz6A-_NATIVE_PAIR_7_57_BARCODE_R51H_1845_25_1 using rosetta_beta version 552
-----
Why two starts? The two messages are consecutive.
Thought for a horrible second that Rosie had me as a dual core but saw that this WU is the only one I had remaining.
ID: 2886 · Report as offensive    Reply Quote
1 · 2 · 3 · Next

Message boards : RALPH@home bug list : Bug reports for Ralph 5.52-5.54



©2024 University of Washington
http://www.bakerlab.org