Posts by Purple Rabbit

1) Message boards : Number crunching : Rosetta Mini with new score terms bug thread (Message 56615)
Posted 2 Nov 2008 by Profile Purple Rabbit
Post:
OK, I'm one for 26. The Linux one amazingly completed correctly on Tomato (host 282106).

This looks like a Windows problem. My other Linux failures may be due to a memory defect...sigh. Onion (host 333136) has been having memory problems so I can't blame Rosetta. The Windows machines are dropping like flies on this app.
2) Message boards : Number crunching : Rosetta Mini with new score terms bug thread (Message 56601)
Posted 1 Nov 2008 by Profile Purple Rabbit
Post:
I've had 100% failure for this app (25 of them so far). I've started aborting them as they appear. This is with BOINC 5.10.45 on both Linux and Windows (XP and Vista), but mainly Windows. I've got one task under Linux that's 4 hours into its 6 hour run. I'll wait to see how that goes.

There's something wrong here (as if I'm telling you something new)!
3) Message boards : Number crunching : Problems with Rosetta version 5.80 (Message 47230)
Posted 29 Sep 2007 by Profile Purple Rabbit
Post:
...I've got 5 more of these puppies waiting...sigh...

An update:

All five of the puppies finished successfully. One of my bad WU was successfully completed by another. The other bad one died
a second time and was put to rest.

With the scattered reports of occasional failures I'm guessing this is probably an initial conditions (random seed) problem
and/or 5.80 not being able to handle the output for particular starting conditions. Things aren't totally broken for "v001",
but something ain't quite right :-)

Rick

Can the forum moderator fix the formatting for this thread? It's way off my screen to the right. I've spent some time
adding hard returns to make my posts more readable.
4) Message boards : Number crunching : Problems with Rosetta version 5.80 (Message 47153)
Posted 28 Sep 2007 by Profile Purple Rabbit
Post:
I think this the same problem Paul is describing.

I've had two WU error out with "error 1" after an hour or two of processing. These are from two
different Linux computers. I had a third one complete successfully on yet another Linux computer. These are the only "v001" WU
I've finished. Looking at Paul's computer he is also seeing some success and some failure with "v001".

All my computers are running Suse 10. The first bad one I shrugged off as gremlins. The second happened less than an hour later
so it looked like a trend :-) I've got 5 more of these puppies waiting...sigh

Bad:

v001_1_NMRREF_1_v001_1_id_model_13IGNORE_THE_REST_idl_2125_1698 Bad #1

v001_1_NMRREF_1_v001_1_id_model_10IGNORE_THE_REST_idl_2125_1167 Bad #2

Good:

v001_1_NMRREF_1_v001_1_id_model_20IGNORE_THE_REST_idl_2125_424 Good #1
5) Message boards : Number crunching : Check pointing needs fixxed (Message 39727)
Posted 22 Apr 2007 by Profile Purple Rabbit
Post:
You can't really keep track of the models via the boinc manager but the rosetta graphics has the model number.

Well yes, but most of my computers are remote Linux machines...sigh. It's a PITA, but certainly not a show stopper. The BOINC progress report is all I have so any efforts to make it more accurate with respect to model completion will be appreciated (but you knew that already!). I only mention this as constructive criticism so you know the problem I have and with the hope that the next version will be better.
6) Message boards : Number crunching : BOINC and electricity (Message 39669)
Posted 20 Apr 2007 by Profile Purple Rabbit
Post:
If I throw the old computer away I will be adding to the landfill. I know they can be recycled, but the closest place to me is 80 miles away and they charge money ($20-$30) for recycling. The landfill is free and 10 miles away. I don't begrudge my 486 its 60 watts. As long as it works it stays out of the landfill.
7) Message boards : Number crunching : Check pointing needs fixxed (Message 39668)
Posted 20 Apr 2007 by Profile Purple Rabbit
Post:
I think one of the unintended consequences of the new user friendly progress bar is that you don't know if a model has completed. I used to wait for 10%, 20%, etc. before doing anything drastic with BOINC. Now these points don't seem to necessarily correspond to the end of a model. My anecdotal and non scientific observations show that I may have thrown away some work because the progress is only updated to reality at the end of each model.

This isn't a complaint. It's an observation. It may also be an example of: "You can't please everyone no matter what you do" :-)
8) Message boards : Number crunching : Problems with Rosetta version 5.59 (Message 39204)
Posted 9 Apr 2007 by Profile Purple Rabbit
Post:
I have had occasional problems with V5.59 on Linux Suse 10.2. Not every result had a problem. The following results (as a sample) died:

http://boinc.bakerlab.org/rosetta/workunit.php?wuid=64664432
http://boinc.bakerlab.org/rosetta/workunit.php?wuid=64280490

More on this machine have died, but not on other machines.

Host 395030 (the one referenced) is a Celeron 1.3 GHz CPU with 640 MB of RAM.

I received a segment fault on these (and other) results. Everything was working OK before 5.59.

My other computers are fine, both Windows and Linux. This seems strange to me. I'm running BOINC 5.8.17 (Windows) and BOINC 5.8.16 (Linux) on my machines. They ought to be current for now.
9) Questions and Answers : Unix/Linux : platform 'x86_64-pc-linux-gnu' not found (Message 30879)
Posted 10 Nov 2006 by Profile Purple Rabbit
Post:
My only point was if the project could easily support the 64 client by mapping it to the 32 bit app then I wanted to encourage them to do so. If not, then end of discussion. I never "demanded" that they support it. It just seemed like an easy thing to do so why not? They may not know this is a problem for a small, but growing group of people.

If nothing else then a link to get the 32 bit app would be nice so I can use the new version with my updated app_info.xml file. I think this is also a simple request, but if the project doesn't want to, end of discussion.

I have a working solution, but it's a kludge. A little help was all I asked. Running the 32 bit BOINC client wasn't appealing to me because I run projects that do have a 64 bit app.

The source code is "official". The compiler makes it work on a specific machine. The project supports specific machines. They can pick and choose. I only wished to encourage them to support mine!

I'd like to have your Z80 version :-) My Heathkit H89 hasn't done anything useful in 18 years. Does it run with 64KB RAM? BTW mine is a 2 MHz Z80 with 5 floppies totaling 1.6 MB of storage (minus 24 KB for Heath DOS, but I can handle CPM too).

I'm done. Both of us have made our points. End transmission :-)

10) Questions and Answers : Unix/Linux : platform 'x86_64-pc-linux-gnu' not found (Message 30749)
Posted 7 Nov 2006 by Profile Purple Rabbit
Post:
My main point is that BOINC allows a number of platform types including 64 bit platforms. The 64 bit platforms are not currently supported by many projects and may require work arounds (app_info.xml file) or a compilation of an executable. Where I get the 64 bit BOINC client is irrelevant. It's open source and the BOINC architecture supports it.

Although a project is not required to support the extended platform format, it's almost trivial for the Rosetta server to map the "x86_64-unknown-Linux-gnu" or whatever platform to the 32 bit app. They should do this. It may not work for all Linux distros, but it would be a good start. The distros that don't work should be documented (by someone, somewhere--the project doesn't need this task)

From a user's point of view, Rosetta should have a place where one can download the current app. This beats having two machines (one 32 bit) to get the app. Having to scrounge it from another machine or doing what One Norse did is beyond what a normal user should have to do. If they can map the 64 bit app requests then that would be great.

Other projects do this successfully. Rosetta can too. What's the big deal? I'm using Suse 10.1 Linux x86-64 so my mileage may differ from yours.
11) Questions and Answers : Unix/Linux : platform 'x86_64-pc-linux-gnu' not found (Message 30622)
Posted 5 Nov 2006 by Profile Purple Rabbit
Post:
I ended up 'solving' the problem for now by downloading the 32 bit boinc client app... modifying the startup and configuration files that were installed by the Debian boinc installer.

Gaaaah! Did you do all this just to get the 32 bit app? I "complained" earlier about not having direct download access to the Rosetta apps. This is really needed until a fix for the 64 bit clients is implemented.

I agree again. Someone at Rosetta needs to make a server change to handle 64 bit app requests.
12) Questions and Answers : Unix/Linux : platform 'x86_64-pc-linux-gnu' not found (Message 30619)
Posted 4 Nov 2006 by Profile Purple Rabbit
Post:
BOINC clients (official or otherwise) advertise their capability by the platform ID. It's up to the server to respond appropriately. Project servers map this ID to the appropriate executable. Some project servers (like Rosetta) don't do this for 64 bit clients. It probably hasn't occurred to them that they need to do this for 64 bit clients. I suspect doing this on the client side is difficult. That's not the BOINC architecture.

Here's my app_info.xml file. It works for me. Of course you have to find the Rosetta executable somewhere. As I said before I got mine from one of my 32 bit Linux computers. You didn't say which Linux you have, but Suse 10.1 (64 bit) runs just fine without wrappers.

<app_info>
<app>
<name>rosetta</name>
</app>
<file_info>
<name>rosetta_5.36_i686-pc-linux-gnu</name>
</file_info>
<app_version>
<app_name>rosetta</app_name>
<version_num>536</version_num>
<file_ref>
<file_name>rosetta_5.36_i686-pc-linux-gnu</file_name>
<main_program/>
</file_ref>
</app_version>
</app_info>
13) Questions and Answers : Unix/Linux : platform 'x86_64-pc-linux-gnu' not found (Message 29997)
Posted 25 Oct 2006 by Profile Purple Rabbit
Post:
How about doing a filter on the request so that instead of an error message, we AMD 64 users automatically get the 32 bit Linux app instead, at least until such time as a 64 bit app is available?

I agree. I just put my new AMD64 3800+ X2 (Suse 10.1 x86-64) on Rosetta. Because it reports being platform type "x86_64-unknown-linux-gnu" it got the error and wouldn't run. The only way to get work was to add an app_info.xml file and "steal" the 32 bit app from one of my other Linux boxes. It runs fine this way except...

Changing the app to a new version is a real pain using this method. First I have to notice that a change is needed, wait for the existing work to clear (or abort them), then make the change. I would like to be able to automatically get the 32 bit app. Also, I haven't found a place to download the app directly. I would have been out of luck if I didn't have another Linux system.
14) Message boards : Number crunching : Report Problems with Rosetta Version 5.25 (Message 20377)
Posted 17 Jul 2006 by Profile Purple Rabbit
Post:
A t354 error (signal 11) on Linux.

http://boinc.bakerlab.org/rosetta/result.php?resultid=28762369

Next Rosetta WU failed to start after error. Restarting BOINC fixed that problem. Running 100% Rosetta (CPDN suspended) on this machine (HT).
15) Message boards : Number crunching : Separate Rosetta Preferences Disappeared (Message 19941)
Posted 9 Jul 2006 by Profile Purple Rabbit
Post:
I had the same problem. It was BAM. It looks like BAM doesn't handle non standard fields properly yet. It also changed/deleted some of my default settings on several projects. I think Willy sort of fixed that one though. I've detached all my computers from BAM for a while. It needs a bit more time to mature.
16) Message boards : Number crunching : Report stuck & aborted WU here please - II (Message 13509)
Posted 12 Apr 2006 by Profile Purple Rabbit
Post:
This one ran for 6 hours stuck at 1.04%. I restarted BOINC and the WU began again at zero. It quickly ran up to 1.04%, but seemed to have hung again according to the graphics display. I aborted the WU after 14 minutes (the second time).

TRUNCATE_TERMINI_FULLRELAX_1enh__433_53_0 using rosetta version 498
http://boinc.bakerlab.org/rosetta/workunit.php?wuid=13904970
17) Message boards : Number crunching : Cancelled work? (Message 12660)
Posted 25 Mar 2006 by Profile Purple Rabbit
Post:
I see both of us have received credit for the cancelled WU's. Thank you.
18) Message boards : Number crunching : Cancelled work? (Message 12651)
Posted 25 Mar 2006 by Profile Purple Rabbit
Post:
I'm seeing the same thing for WUs:

http://boinc.bakerlab.org/rosetta/workunit.php?wuid=12023676
http://boinc.bakerlab.org/rosetta/workunit.php?wuid=12002192
http://boinc.bakerlab.org/rosetta/workunit.php?wuid=11965631

I've got two more within minutes of completion so I haven't aborted them. I checked the rest of my list and I don't have any more (yet). I have the same question since no notice was given for this.
19) Message boards : Number crunching : Any Xtra Credit for Protein of the Day? (Message 12344)
Posted 20 Mar 2006 by Profile Purple Rabbit
Post:
No extra cobblestones, but how about a sequence of colored protein strands (or some symbol), one for each time up to something like four, next to the person's name in the forums? I would differ with River in that this would persist. I suspect this might be easier to manage from a project standpoint instead of deleting "extra credit points" every so often.

If not that, then it's only worth an "atta boy". This "award" is really the luck of the draw and doesn't reflect anyone's true efforts.

I'm not sure that a footnote citing "Purple Rabbit" would be academically acceptable :-)
20) Message boards : Number crunching : Reporting Results Anomaly (Message 12338)
Posted 20 Mar 2006 by Profile Purple Rabbit
Post:
Occasionally (for the past few weeks when I manually report a result) the connection to the scheduler succeeds, but the result is not reported to the project as shown by the project results web page. I haven't kept notes, but I think this has happened about 5 times. The majority of the time everything works as it is supposed to.

The message file shows a 3 minute delay (timeout?) between requesting the report and a succeed message from the server.

I'm guessing there can be a delay between the upload server and the scheduler. Maybe the scheduler doesn't know that I uploaded a result? This happens (sometimes) when I try to immediately report an uploaded result. Shouldn't I get a different message if the result isn't received correctly?

This is a non-problem since everything works eventually. The next time BOINC reports the result it is received.

I only mention it since it might indicate a communications bottleneck or a timing (load) problem between the servers.

"Doctor, it hurts when I do this."
"Then stop doing it" :-)


Next 20



©2022 University of Washington
https://www.bakerlab.org