Message boards : Number crunching : Problems and Technical Issues with Rosetta@home
Previous · 1 . . . 288 · 289 · 290 · 291 · 292 · 293 · 294 . . . 295 · Next
Author | Message |
---|---|
Sid Celery Send message Joined: 11 Feb 08 Posts: 2082 Credit: 40,621,050 RAC: 4,944 |
Wow, the queue got eaten up that fast? Are you sure? I didn't even see a queue appear on the front page, just downloaded 16 (1 per core) 8hrs ago, then nothing, then 8 about 5hrs ago, then nothing again. Weird thing is, 10 of that first 16 Robetta tasks crashed out after varying amounts of times when they were pretty reliable before. So, just 14 running successfully atm with nothing to back it up. Edit: Just checked and they were mainly re-sends, very few new tasks. The ones that crashed out had already crashed on a previous host, so not a sign of anything new going wrong |
Grant (SSSF) Send message Joined: 28 Mar 20 Posts: 1635 Credit: 16,775,951 RAC: 11,387 |
No, the problem is the database is corrupted.I've set the messages boards to ignore pinned messages. It should include this thread in the list in order of most recent post to oldest (as per my forum settings), and for some reason, that's no longer occurring. Edit- a "Rosetta Beta 6.00" is borked as well, it's not being sorted in order of post date/time. Edit- i thought i'd see what changing the settings fir the message boards would do- and for some reason the how to sort option for the threads had changed from "Newest post first" to "Newest first." Changed it back and updated- threads are now all being sorted as they should be. If weird things start happening, check the settings (even if you haven't made any changes to them for years...). Grant Darwin NT |
[VENETO] boboviz Send message Joined: 1 Dec 05 Posts: 1980 Credit: 9,197,551 RAC: 2,407 |
P.S. No problem :-P I thought he was asking how to block the Beta app so he wouldn't get that work, being that it errors out right now. With the xml file you force to download the selected app. |
dcs1955 Send message Joined: 2 Dec 22 Posts: 13 Credit: 5,322,205 RAC: 8,940 |
Since the project has recently restarted I have had some tasks start with a predicted exec. time of 10hr. After crunching for 8 hrs it then says that there are 15-24 more hrs. I am running windows 11. Is this a false estimate, or will it take that much time? To not pour more time into something acting weird I have aborted the task. Is this a problem? Will it be worth crunching? Will I get full credit for all this processing? |
LigH Send message Joined: 7 Sep 09 Posts: 25 Credit: 9,241,214 RAC: 10 |
I don't receive new tasks for weeks now. Is there a systematic list of checks I should do to discover the reason why? There used to be a time when I did not get new tasks because I had to enable Python in my web profile, I did that once. Has there been a similar change which expects me to change any preferences which I never heard of before? Fun and success! Jobs: holzon + 12angebote Hobbies: doom9/Gleitz + PlaneShift |
Grant (SSSF) Send message Joined: 28 Mar 20 Posts: 1635 Credit: 16,775,951 RAC: 11,387 |
Since the project has recently restarted I have had some tasks start with a predicted exec. time of 10hr. After crunching for 8 hrs it then says that there are 15-24 more hrs. I am running windows 11. Is this a false estimate, or will it take that much time? To not pour more time into something acting weird I have aborted the task. Is this a problem? Will it be worth crunching? Will I get full credit for all this processing?It does take 8 hours to process each Task, but as your system is severely over committed, it's taking it 12 hours to do 8 hours worth of work. You could determine what else is using your CPU time (eg Folding@home) and reduce the number of cores/threads available to BOINC to sort the issue out, or just accept that it takes your system 12 hours to do only 8 hours worth of work. Grant Darwin NT |
Grant (SSSF) Send message Joined: 28 Mar 20 Posts: 1635 Credit: 16,775,951 RAC: 11,387 |
I don't receive new tasks for weeks now.Because the servers were down for several weeks- as was mentioned in this thread earlier. And even before that outage, work availability was very much random. Sometimes there was work, often there wasn't. Grant Darwin NT |
LigH Send message Joined: 7 Sep 09 Posts: 25 Credit: 9,241,214 RAC: 10 |
Aaaand ... I just received new tasks. Fun and success! Jobs: holzon + 12angebote Hobbies: doom9/Gleitz + PlaneShift |
Greg_BE Send message Joined: 30 May 06 Posts: 5690 Credit: 5,859,226 RAC: 8 |
Current status: Tasks ready to send 0 Tasks in progress 66224 |
hadron Send message Joined: 4 Sep 22 Posts: 68 Credit: 1,543,060 RAC: 269 |
You can ask Veneto how to do that modification to block beta work. But I only got one beta and since then nothing but clean 4.20. So is it worth the effort to mess around? This won't block Boinc from downloading beta work, and it has no effect all all on the number of cores each task will use.. It will only limit it from running more than X Rosetta tasks simultaneously. To control the number of cores a task uses, you need to include an <app_version> section in app_config.xml, like this: <app_version> <app_name>ATLAS</app_name> <avg_ncpus>2</avg_ncpus> <plan_class>vbox64_mt_mcore_atlas</plan_class> <cmdline>--nthreads 2</cmdline> </app_version> <avg_ncpus> tells Boinc that tasks will be running on 2 threads, while the cmdline parameter is passed to the program to tell it to do so. If there were any way to tell Boinc not to download beta tasks, it would be in the cc_config.xml file. I see nothing in the user guide to suggest that is even possible. Therefore, since Rosetta itself does not distinguish between the "old' and beta tasks, it seems there is no way to block a client Boinc from downloading Rosetta beta tasks. |
hadron Send message Joined: 4 Sep 22 Posts: 68 Credit: 1,543,060 RAC: 269 |
Stuprd boinc thinks <max_concurrent>0</max_concurrent> is unlimited. How would you do it? |
kotenok2000 Send message Joined: 22 Feb 11 Posts: 257 Credit: 476,390 RAC: 533 |
-1 |
hadron Send message Joined: 4 Sep 22 Posts: 68 Credit: 1,543,060 RAC: 269 |
If you want to complain about how Boinc works, then you owe it to everyone to suggest an alternative. So once again, how would you do it? |
kotenok2000 Send message Joined: 22 Feb 11 Posts: 257 Credit: 476,390 RAC: 533 |
<max_concurrent>-1</max_concurrent> should have been unlimited. |
hadron Send message Joined: 4 Sep 22 Posts: 68 Credit: 1,543,060 RAC: 269 |
<max_concurrent>-1</max_concurrent> should have been unlimited. Alternatively, don't even include the section at all, if you don't wish to limit the number of concurrent tasks. |
Grant (SSSF) Send message Joined: 28 Mar 20 Posts: 1635 Credit: 16,775,951 RAC: 11,387 |
boinc-process is still dead (days), along with Ralph (week and a half and counting). There was another very brief release of Rosetta 4.20 Tasks a few hours ago, so a bit more to go on the Validation backlog. Grant Darwin NT |
PMH_UK Send message Joined: 9 Aug 08 Posts: 15 Credit: 1,243,749 RAC: 0 |
Ralph is alive but a bit unwell. Paul. |
[VENETO] boboviz Send message Joined: 1 Dec 05 Posts: 1980 Credit: 9,197,551 RAC: 2,407 |
Ralph is alive but a bit unwell. Yep 23/08/2024 18:16:13 | ralph@home | Scheduler request to https://ralph.bakerlab.org/ralph_cgi/cgi failed: HTTP file not found |
Grant (SSSF) Send message Joined: 28 Mar 20 Posts: 1635 Credit: 16,775,951 RAC: 11,387 |
Ralph is alive but a bit unwell.And clicking on Number Crunching there gives this- Deprecated: trim(): Passing null to parameter #1 ($string) of type string is deprecated in /projects/boinc/ralph/html/inc/util_basic.inc on line 107Followed by the thread list. So it looks like it's back to where it was, just before it died completely. I wouldn't say it was so much alive, as it's no longer completely dead. Just mostly dead. Grant Darwin NT |
Grant (SSSF) Send message Joined: 28 Mar 20 Posts: 1635 Credit: 16,775,951 RAC: 11,387 |
Ralph appears to be fully functional now, so with a bit of luck we might get some new work there to test Rosetta 4.29 (and maybe even the Beta 6.06 application as well that crashed and burned everything that was sent out for it on it's fist try here). While we're doing that, they can sort out the dead boinc-process server here. Grant Darwin NT |
Message boards :
Number crunching :
Problems and Technical Issues with Rosetta@home
©2024 University of Washington
https://www.bakerlab.org