Forums :
Technical Support :
Test work?
Message board moderation
Previous · 1 · 2 · 3 · Next
| Author | Message |
|---|---|
 Stefan LedwinaVolunteer tester  Send message Joined: 7 Jun 07 Posts: 47 Credit: 4,696,550 RAC: 0 |
Guess what? The option to allow test workunits appeared! =) http://www.cosmologyathome.org/prefs.php?subset=project [edit] But I still get - 06.09.2008 20:04:22|Cosmology@Home|Message from server: No work sent 06.09.2008 20:04:22|Cosmology@Home|Message from server: (there was work but it was committed to other platforms) LoL  last.fm BOINC users group 
|
|
Brian Silvers Send message Joined: 11 Dec 07 Posts: 420 Credit: 270,580 RAC: 0 |
Guess what? Yes, I see the option is available now. I don\'t know why it didn\'t dawn on me that I had downloaded \"retreads\". I was real tired at the time... Anyway, I suspect it takes at least two connection attempts to even be requesting the test work, but I could be wrong. What I\'m thinking is the first connection just asks for 2.15 work, but it does get the flag that you\'re allowing test work. The second request would then ask for both \"normal\" and \"test\". All of what I just said is null-and-void if the test flag is server-side only though... :-) I think what we\'re seeing is that they ran dry on test units, but still have those retreads... Dunno... 
|
|
Jord Volunteer moderator Volunteer tester  Send message Joined: 15 Jun 07 Posts: 345 Credit: 50,500 RAC: 0 |
Guess what? With thanks to Scott for making the necessary changes. A simple line he forgot to add to an .inc file. In the mean time we\'re still using HR, more coarse on the test tasks than on the normal tasks, but it still needing to be available for your platform. My AMD did get work earlier (normal 2.15), my Intel has been dry all the time. We\'ll probably have to wait until the top of the queue is done. I wouldn\'t be surprised if those are still old tasks with the old credit. |
caspr Send message Joined: 8 Aug 07 Posts: 54 Credit: 527,780 RAC: 0 |
Guess what? Even with test apps. enabled I\'m still getting the \"committed to other platforms message\" BTW,...This is on a windoz 64bit os A clear conscience is usually the sign of a bad memory |
|
Brian Silvers Send message Joined: 11 Dec 07 Posts: 420 Credit: 270,580 RAC: 0 |
Amazingly enough, my AMD finally got 4 \"brand new\" tasks! No, not the test work, but new, meaning that there are only 2 replications and mine were all the 2nd replication. The work units were all \"wu_081308...\" This probably means that the generator was working all this time, but as someone theorized, the older work (\"retreads\") took priority...?
|
|
Jayargh Volunteer moderator Volunteer tester  Send message Joined: 25 Jun 07 Posts: 508 Credit: 2,282,158 RAC: 0 |
If you are running the \'Old Web Design\' you will not see the option of run test apps....only the main web site displays this at the moment for anyone who wants to test ;) |
|
rbpeake Send message Joined: 27 Jun 07 Posts: 118 Credit: 61,883 RAC: 0 |
Lots of test work now, the work units seem to be flowing through in quantity! :) |
|
Phoneman1 Send message Joined: 5 Nov 07 Posts: 113 Credit: 3,100,327 RAC: 0 |
After seeing Stefan\'s post last night I went back and checked that page - the test question had indeeed appeared so I set it yes. However, that did not result in any more test units coming my way immmediately, so I left it to it overnight. Checking the log this morning I see the last \"committed to other platforms\" message was at 22:53:07 utc. The next schedule request (on the other machine at 22:54:00 utc) got test units. Since that time all requests for a positve number of seconds of work have been met with test work units. - I guess something must have been changed on the server around then (note to the developers I think the server time is a few minutes out - I am quoting times from my logs and converting to utc). I can also report that of the 12 test units completed to date 5 have already been completed by the wingmen and 140 credits have been awarded for each. So all in all looking pretty good from here at the moment, fingers firmly crossed! Phoneman1 |
|
AnRM Send message Joined: 3 Sep 07 Posts: 19 Credit: 2,255,870 RAC: 0 |
We have reattached our boxes without any apparent problems....good to be back! Cheers, Rog. |
|
Phoneman1 Send message Joined: 5 Nov 07 Posts: 113 Credit: 3,100,327 RAC: 0 |
I\'ve just noticed this in the logs for one machine(times are utc +1hr): 07-Sep-2008 07:27:32 [Cosmology@Home] Started upload of wu_test_090608_121927_0_0_0 The HTTP internal server error has been seen on numerous times over the last few months. I don\'t recall seeing the already reported message very often though. The HTTP message is often associated with an assigned work unit(s) that isn\'t / aren\'t downloaded. I\'ve checked my task list on the web and found wu_test_090608_175744_0 assigned to this machine but it has not been downloaded (no mention of it in the logs either). So the HTTP internal error remains an issue I think. It doesn\'t seem to do much harm if it is an infrequent occurrence as a re-send will be triggered after 10 days. Incidentally, I notice that my Intel Vista machine is now sometimes being paired with an AMD running XP. Hope that is OK. Phoneman1 |
|
Phoneman1 Send message Joined: 5 Nov 07 Posts: 113 Credit: 3,100,327 RAC: 0 |
The dreaded \"committed to other platform\" messages have re-appeared since 18:31:34 utc. Boinc made automatic repeated requests for work and a total of eight such mesages were received. Not sure why that happened all of a sudden, I haven\'t changed any settings. A bit later at 18:53:14 utc a regular, non test, work unit downloaded. A bit later still at 19:07:42 utc a batch of 2 test units and one regular unit was downloaded. There follwed another three instances of the \"committed to other platform\" messages, followed by three futher downloads of one test unit on each request. These latest test units have an id with test_090608_ indicating they were created yesterday - before the 18:31:34 message mentioned above. I am not sure why they weren\'t downloaded at that request (they are suffix 1 and 0 and so aren\'t resends). Other than that there has been nothing obviously amiss since my earlier post so that has been 11.5 hours of trouble free running today. One quad was running test Cosmology units only and the other quad spent most of the day running QMC and Cosmology test units until the QMC work was finished. The input queue of work has remained pretty constant, hovering around the 3 days I like to maintain. Phoneman1 |
|
Brian Silvers Send message Joined: 11 Dec 07 Posts: 420 Credit: 270,580 RAC: 0 |
The dreaded \"committed to other platform\" messages have re-appeared since 18:31:34 utc. Boinc made automatic repeated requests for work and a total of eight such mesages were received. Not sure why that happened all of a sudden, I haven\'t changed any settings. I think it\'s that the test units ran dry, then all that was left were the \"retreads\"... On that note, as we were not told which to give priority to, I\'ve been giving priority to the \"retreads\" or the 2.15 work that I had already picked up.
|
|
Phoneman1 Send message Joined: 5 Nov 07 Posts: 113 Credit: 3,100,327 RAC: 0 |
I agree the test units do seem to have run dry, no further test ones or reglular ones have downloaded in the intervening hours. It is still a puzzle to me why I got two successful requests for downloads after the first eight \"commited to other platform\" messages yesterday eveneing. The test units downloaded weren\'t re-sends as they were suffix _0 and _1 types, the regular ones were re-sends. It may be that the test units weren\'t assinged to any HR pool and had to be assigned to a pool but there was a gap of over 35 minutes between the first \"committed to other platform\" message and them downloading on the nineth attempt. It looks to me as if there could still be an issue somewhere with HR and the feeder mechanism and it would be worth the developers checking the server logs between 18:20and 19:20 utc (server time is slightly adrift so that hour should cover the events). Phoneman1 |
Thunder Send message Joined: 15 Apr 08 Posts: 101 Credit: 4,535,998 RAC: 0 |
So it looks like a majority of the test units have come back home to roost. Do we know anything more about whether the coarse HR is going to work instead of fine HR? 
|
Conan Send message Joined: 28 Aug 07 Posts: 169 Credit: 1,262,905 RAC: 236 |
I don\'t think it seems to matter what type of work unit they are as none or at least very few are being downloaded. Currently over 79,000 work units available for download only 1700 are in progress. I can\'t get any at all, on either Windows or Linux. Just getting \"committed to other platforms\". I am interested as which platform is getting any work. I stopped Cosmo due to huge amount of zero credit that was granted to me on well over 200 hours of work with the casual cancelling of all those WU\'s and the statement that we had been given lots of credit in the past. The past is not now, I was not impressed at that flippant attitude. I have restarted Cosmo in the hope that things are working again, that the owner of this project has taken some more responsibility in running it and to see what the credit issue is like. I will run a few and see what I get, I would like to get to my million on this project. |
|
Conan Send message Joined: 28 Aug 07 Posts: 169 Credit: 1,262,905 RAC: 236 |
I don\'t think it seems to matter what type of work unit they are as none or at least very few are being downloaded. Still over 79,000 work units to send and now only 1,400 are being processed. Test work units or otherwise no one is getting work, come on project team, kick that server into gear. |
|
Brian Silvers Send message Joined: 11 Dec 07 Posts: 420 Credit: 270,580 RAC: 0 |
It looks like, as happened before, someone from the project got something running (thanks for that!), but then they wandered off and didn\'t make sure things didn\'t get bogged down... :sigh: Since Burdett is in this thread, I won\'t feel so bad in picking on them... ;-) If the status page is to be believed and the test results are included in the pool of results in progress, Burdett would own approximately 5-6% of the tasks in progress just by themselves on one host. Also, sorry Burdett, they have a few of the \"test\" WUs where they were their own wingman, although on another host. Yes, Burdett, I know all of that\'s not your fault, other than perhaps too large of a cache that grabbed a lot of test tasks, but whatever... The project should\'ve put restrictions on how many test tasks a single host could get so that more combinations could be tested... This is something common to this project though: short-sightedness... :( We\'re clearly at a logjam again...
|
|
.clair. Send message Joined: 4 Nov 07 Posts: 604 Credit: 10,881,302 RAC: 19 |
I like the reason for cosmo to exist but It is the persistent problems that stop me from crunching this project i just keep reading and waiting for things to smooth out. |
|
Klimax Send message Joined: 24 Oct 07 Posts: 22 Credit: 648,291 RAC: 0 |
Well,just four hours ago I got some test units (12). |
|
Thunder Send message Joined: 15 Apr 08 Posts: 101 Credit: 4,535,998 RAC: 0 |
Yeah, I have test work with today\'s date, so I *guess* they\'re doing a second round of testing. It\'s either that or they threw some more work in just so folks would stop complaining about not having any. Of course if they\'d just post something we\'d know which is which. :P
|
