Ad@m\'s BLOG (page 3)


Advanced search

Message boards : SZTAKI Desktop Grid : Ad@m\'s BLOG (page 3)

AuthorMessage
Profile kadam
Project administrator
Avatar
Send message
Joined: May 25 05
Posts: 589
Credit: 38,614
RAC: 0
Message 1650 - Posted 8 Feb 2006 17:19:49 UTC

    Most likely the venue problem seems to be gone (hopefully forever). I do appologize for the long delay, but I really couldn\'t find a clue what could be wrong. Now today I\'ve been fighting with the database, remerged everything again (it was a pain in the ass...) and now this time it seemed to work...

    I kindly ask everyone to check the clients and report anything unusal...

    A nice occasion to open a new page... ;]

    Everyone! Let\'s get back to crunching...!
    ____________
    If you like BOINC, you may also find CaretCursor to be appealing.

    Profile Paul D. Buck
    Send message
    Joined: Jul 20 05
    Posts: 35
    Credit: 57,732
    RAC: 0
    Message 1655 - Posted 8 Feb 2006 18:06:58 UTC

      Last modified: 8 Feb 2006 18:09:08 UTC

      The lesson is to tell us something even if negative. Lists of things you have tried ... printouts of the table sturcture to show us idiots that you looked at it ...

      And I reattached one machine ... more to follow ...

      ==== edit
      Oh, and SDG is using the correct venue ... very strange ... :)

      Sorry, only sense of humor I have ...
      ____________

      Profile kadam
      Project administrator
      Avatar
      Send message
      Joined: May 25 05
      Posts: 589
      Credit: 38,614
      RAC: 0
      Message 1656 - Posted 8 Feb 2006 18:08:52 UTC - in response to Message 1655.

        The lesson is to tell us something even if negative


        Lesson learned...now let\'s chrunch some chrunchies...
        ____________
        If you like BOINC, you may also find CaretCursor to be appealing.

        Profile UBT - Halifax--lad
        Avatar
        Send message
        Joined: Sep 10 05
        Posts: 126
        Credit: 3,147
        RAC: 0
        Message 1661 - Posted 8 Feb 2006 20:10:22 UTC - in response to Message 1656.

          The lesson is to tell us something even if negative


          Lesson learned...now let\'s chrunch some chrunchies...


          So your sticking around now rather than going missing for so long?
          ____________
          Join us in Chat (see the forum) Click the Sig


          Join UBT

          Profile kadam
          Project administrator
          Avatar
          Send message
          Joined: May 25 05
          Posts: 589
          Credit: 38,614
          RAC: 0
          Message 1663 - Posted 8 Feb 2006 20:11:14 UTC - in response to Message 1661.

            So your sticking around now rather than going missing for so long?


            Sounds like a plan...
            ____________
            If you like BOINC, you may also find CaretCursor to be appealing.

            Profile UBT - Halifax--lad
            Avatar
            Send message
            Joined: Sep 10 05
            Posts: 126
            Credit: 3,147
            RAC: 0
            Message 1665 - Posted 8 Feb 2006 20:30:16 UTC - in response to Message 1663.

              So your sticking around now rather than going missing for so long?


              Sounds like a plan...


              Thats good then!
              ____________
              Join us in Chat (see the forum) Click the Sig


              Join UBT

              Profile Nightbird
              Forum moderator
              Avatar
              Send message
              Joined: Jul 12 05
              Posts: 920
              Credit: 114,924
              RAC: 0
              Message 1666 - Posted 8 Feb 2006 20:41:12 UTC - in response to Message 1661.

                Last modified: 8 Feb 2006 20:42:07 UTC

                So your sticking around now rather than going missing for so long?
                Sounds like a plan...

                Adam, do you know that The Communication is important with your crunchers ?
                We need to be in communication with you and you with us !

                ps : Where is Attila ? disappeared ?

                ____________

                Profile kadam
                Project administrator
                Avatar
                Send message
                Joined: May 25 05
                Posts: 589
                Credit: 38,614
                RAC: 0
                Message 1667 - Posted 8 Feb 2006 21:15:58 UTC - in response to Message 1666.

                  Last modified: 8 Feb 2006 21:16:05 UTC


                  Adam, do you know that The Communication is important with your crunchers ?
                  We need to be in communication with you and you with us !

                  I know, but I hate to communicate bad news again and again...it won\'t happen again, sry


                  ps : Where is Attila ? disappeared ?

                  Seems like...

                  ____________
                  If you like BOINC, you may also find CaretCursor to be appealing.

                  Profile Nightbird
                  Forum moderator
                  Avatar
                  Send message
                  Joined: Jul 12 05
                  Posts: 920
                  Credit: 114,924
                  RAC: 0
                  Message 1668 - Posted 8 Feb 2006 21:32:50 UTC - in response to Message 1667.


                    Adam, do you know that The Communication is important with your crunchers ?
                    We need to be in communication with you and you with us !

                    I know, but I hate to communicate bad news again and again...it won\'t happen again, sry


                    ps : Where is Attila ? disappeared ?

                    Seems like...

                    Well, if you\'re alone, that can explain some things. :(




                    ____________

                    Profile Nightbird
                    Forum moderator
                    Avatar
                    Send message
                    Joined: Jul 12 05
                    Posts: 920
                    Credit: 114,924
                    RAC: 0
                    Message 1672 - Posted 9 Feb 2006 1:15:23 UTC

                      Last modified: 9 Feb 2006 1:22:30 UTC

                      Your fix seems definitively ok but something is amazing : that\'s easier now to get wus from the server !
                      Before when we saw \"Ready to send < 100\" and even with the master program running, often we got this message \"No work sent ; There was work but it was committed to others platforms\".
                      But for now, we get wus from the server.
                      Are we lucky ?
                      ____________

                      Profile kadam
                      Project administrator
                      Avatar
                      Send message
                      Joined: May 25 05
                      Posts: 589
                      Credit: 38,614
                      RAC: 0
                      Message 1689 - Posted 9 Feb 2006 12:49:08 UTC - in response to Message 1672.

                        Well, Nightbird, it\'s not that big luck.
                        After I\'ve managed to solve the venue problem, all the errors have been gone except for the \'there was work, but it was committed to other platform\'. Than I\'ve realized that it\'s not an error it\'s just the drawback of homogeneous redunacy, which I\'ve turned on to have less false positive results. But we\'ve had a meeting with the mathematicians last week and they\'ve told me that it\'s not a big deal to sort out those few false positive results...so homogeneous redundancy is turned off and workunits are flowing smoother now...
                        ____________
                        If you like BOINC, you may also find CaretCursor to be appealing.

                        Profile Nightbird
                        Forum moderator
                        Avatar
                        Send message
                        Joined: Jul 12 05
                        Posts: 920
                        Credit: 114,924
                        RAC: 0
                        Message 1693 - Posted 9 Feb 2006 13:34:04 UTC

                          Last modified: 9 Feb 2006 13:35:13 UTC

                          Lol
                          I would say \"enfer et damnation\"
                          and \"your\" answer was here also
                          (see Lee Carre)
                          http://szdg.lpds.sztaki.hu/szdg/forum_thread.php?id=176#1460)

                          ____________

                          Ageless
                          Avatar
                          Send message
                          Joined: Jul 12 05
                          Posts: 64
                          Credit: 3,001
                          RAC: 0
                          Message 1696 - Posted 10 Feb 2006 7:00:12 UTC

                            Last modified: 10 Feb 2006 7:00:29 UTC

                            When editing General preferences, double positive and negative:

                            Disconnect when done?
                            (matters only if you use a modem) yesyes/nono
                            ____________
                            Jord - "Looking for an answer solves 90% of the questions asked!"

                            Boinc Wiki - Information at your fingertips.

                            Profile kadam
                            Project administrator
                            Avatar
                            Send message
                            Joined: May 25 05
                            Posts: 589
                            Credit: 38,614
                            RAC: 0
                            Message 1697 - Posted 10 Feb 2006 8:10:17 UTC - in response to Message 1696.

                              Last modified: 10 Feb 2006 12:01:32 UTC

                              Thanks for the note, Ageless! I\'ve had it corrected.
                              ____________
                              If you like BOINC, you may also find CaretCursor to be appealing.

                              Profile Ananas
                              Send message
                              Joined: Jul 12 05
                              Posts: 222
                              Credit: 665,833
                              RAC: 0
                              Message 1743 - Posted 12 Feb 2006 16:05:02 UTC

                                Oh no, it\'s back !!!

                                SZTAKI Desktop Grid - 2006-02-12 17:19:00 - Message from server: No work sent (there was work but it was committed to other platforms)

                                Profile Nightbird
                                Forum moderator
                                Avatar
                                Send message
                                Joined: Jul 12 05
                                Posts: 920
                                Credit: 114,924
                                RAC: 0
                                Message 1745 - Posted 12 Feb 2006 17:56:06 UTC - in response to Message 1743.

                                  Last modified: 12 Feb 2006 17:58:15 UTC

                                  Oh no, it\'s back !!!

                                  SZTAKI Desktop Grid - 2006-02-12 17:19:00 - Message from server: No work sent (there was work but it was committed to other platforms)

                                  Yes it\'s back since yesterday for me.
                                  I update the project manually and i get wus but the question is :
                                  why and again the problem ??
                                  homogeneous redundancy is turned off ?
                                  ____________

                                  Profile UBT - Halifax--lad
                                  Avatar
                                  Send message
                                  Joined: Sep 10 05
                                  Posts: 126
                                  Credit: 3,147
                                  RAC: 0
                                  Message 1748 - Posted 12 Feb 2006 20:22:02 UTC

                                    Well I hope when Adam gets in on Monday he can find the cause of it.

                                    If not he is best off asking the other projects for help
                                    ____________
                                    Join us in Chat (see the forum) Click the Sig


                                    Join UBT

                                    Profile Nightbird
                                    Forum moderator
                                    Avatar
                                    Send message
                                    Joined: Jul 12 05
                                    Posts: 920
                                    Credit: 114,924
                                    RAC: 0
                                    Message 1749 - Posted 12 Feb 2006 22:14:33 UTC - in response to Message 1748.

                                      Last modified: 12 Feb 2006 22:17:38 UTC

                                      Well I hope when Adam gets in on Monday he can find the cause of it.

                                      If not he is best off asking the other projects for help

                                      Tomorrow, Adam will have \"some\" problems to resolve (wap stats, forum preferences, join a team, add-ons page and no work). ;)

                                      ____________

                                      Profile UBT - Halifax--lad
                                      Avatar
                                      Send message
                                      Joined: Sep 10 05
                                      Posts: 126
                                      Credit: 3,147
                                      RAC: 0
                                      Message 1774 - Posted 14 Feb 2006 19:24:44 UTC - in response to Message 1755.

                                        Attila should keep an eye on the WUs, it seems like he doesn\'t do so...
                                        ...I\'ll be back on friday and solve all the problems mentioned.


                                        give him a slap from us all when you get back

                                        ____________
                                        Join us in Chat (see the forum) Click the Sig


                                        Join UBT

                                        Profile Attila Csaba Marosi
                                        Forum moderator
                                        Project developer
                                        Avatar
                                        Send message
                                        Joined: Jul 12 05
                                        Posts: 76
                                        Credit: 26,363
                                        RAC: 0
                                        Message 1779 - Posted 15 Feb 2006 11:10:55 UTC - in response to Message 1774.

                                          Last modified: 15 Feb 2006 13:35:55 UTC

                                          Attila should keep an eye on the WUs, it seems like he doesn\'t do so...
                                          ...I\'ll be back on friday and solve all the problems mentioned.


                                          give him a slap from us all when you get back


                                          altough i am not sure if adam is refering to me, please don\'t kill
                                          the poor overcommitted sysadmin ;S

                                          attila

                                          ps.: you should have WU-s now

                                          ____________

                                          Ageless
                                          Avatar
                                          Send message
                                          Joined: Jul 12 05
                                          Posts: 64
                                          Credit: 3,001
                                          RAC: 0
                                          Message 1784 - Posted 15 Feb 2006 16:08:43 UTC - in response to Message 1779.

                                            ps.: you should have WU-s now

                                            If only it was for my platform. ;)

                                            15/02/2006 17:13:25|SZTAKI Desktop Grid|Message from server: No work sent
                                            15/02/2006 17:13:25|SZTAKI Desktop Grid|Message from server: (there was work but it was committed to other platforms)
                                            15/02/2006 17:13:25|SZTAKI Desktop Grid|No work from project

                                            ____________
                                            Jord - "Looking for an answer solves 90% of the questions asked!"

                                            Boinc Wiki - Information at your fingertips.

                                            Profile Attila Csaba Marosi
                                            Forum moderator
                                            Project developer
                                            Avatar
                                            Send message
                                            Joined: Jul 12 05
                                            Posts: 76
                                            Credit: 26,363
                                            RAC: 0
                                            Message 1785 - Posted 15 Feb 2006 16:21:09 UTC - in response to Message 1784.

                                              ps.: you should have WU-s now

                                              If only it was for my platform. ;)

                                              15/02/2006 17:13:25|SZTAKI Desktop Grid|Message from server: No work sent
                                              15/02/2006 17:13:25|SZTAKI Desktop Grid|Message from server: (there was work but it was committed to other platforms)
                                              15/02/2006 17:13:25|SZTAKI Desktop Grid|No work from project


                                              try now
                                              ____________

                                              Ageless
                                              Avatar
                                              Send message
                                              Joined: Jul 12 05
                                              Posts: 64
                                              Credit: 3,001
                                              RAC: 0
                                              Message 1787 - Posted 15 Feb 2006 17:01:46 UTC - in response to Message 1785.

                                                Thanks Attila, working now.

                                                ____________
                                                Jord - "Looking for an answer solves 90% of the questions asked!"

                                                Boinc Wiki - Information at your fingertips.

                                                Attalla
                                                Send message
                                                Joined: Sep 19 05
                                                Posts: 56
                                                Credit: 314,131
                                                RAC: 0
                                                Message 1805 - Posted 16 Feb 2006 21:14:59 UTC

                                                  Last modified: 16 Feb 2006 21:16:22 UTC

                                                  Szemelyes fiok:
                                                  Osszes kredit 44,107.64
                                                  Kozelmultbeli teljesitmeny 215.18

                                                  Kozelmultbeli teljesitmeny szerint rendezve:
                                                  27 Attalla 219.44 43,999.03 Hungary 19 Sep 2005 17:34:47 UTC

                                                  Osszes kredit szerint rendezve:
                                                  26 Attalla 213.07 44,006.68 Hungary 19 Sep 2005 17:34:47 UTC

                                                  3 adatbazis?
                                                  ____________

                                                  Profile Ananas
                                                  Send message
                                                  Joined: Jul 12 05
                                                  Posts: 222
                                                  Credit: 665,833
                                                  RAC: 0
                                                  Message 1820 - Posted 17 Feb 2006 0:37:32 UTC

                                                    Last modified: 17 Feb 2006 1:02:05 UTC

                                                    The 503 scheduler produces ghost WUs in all projects that already have it and sometimes it even seems to loose results.

                                                    502 was much better.


                                                    p.s.: I have to correct myself - 502 is already one of those versions with many ghosts, the earlier ones didn\'t have that problem so much.

                                                    Lee Carre
                                                    Send message
                                                    Joined: Sep 30 05
                                                    Posts: 65
                                                    Credit: 7,033
                                                    RAC: 0
                                                    Message 1821 - Posted 17 Feb 2006 1:21:21 UTC - in response to Message 1820.

                                                      The 503 scheduler produces ghost WUs in all projects that already have it and sometimes it even seems to loose results.

                                                      502 was much better.


                                                      p.s.: I have to correct myself - 502 is already one of those versions with many ghosts, the earlier ones didn\'t have that problem so much.

                                                      is the new(est) one any better?
                                                      surely some kind of handshake should be done to check that a client actually has the work?

                                                      Profile Ananas
                                                      Send message
                                                      Joined: Jul 12 05
                                                      Posts: 222
                                                      Credit: 665,833
                                                      RAC: 0
                                                      Message 1824 - Posted 17 Feb 2006 6:41:56 UTC

                                                        Someone told me that Einstein uses a feature that allows the server to tell the client what he missed.

                                                        I think this is not more than a workaround though, the Berkeley guys messed up something and do not find the bug.

                                                        The new feature (when enabled) will not help me anyway, nearly all of my boxes are on 4.19 and the workaround will only affect quite new core clients.

                                                        Lee Carre
                                                        Send message
                                                        Joined: Sep 30 05
                                                        Posts: 65
                                                        Credit: 7,033
                                                        RAC: 0
                                                        Message 1833 - Posted 17 Feb 2006 14:13:00 UTC - in response to Message 1824.

                                                          Someone told me that Einstein uses a feature that allows the server to tell the client what he missed.

                                                          I think this is not more than a workaround though, the Berkeley guys messed up something and do not find the bug.

                                                          The new feature (when enabled) will not help me anyway, nearly all of my boxes are on 4.19 and the workaround will only affect quite new core clients.

                                                          I heard about that, on every scheduler contact it checks that both agree on what work the client has, but apparently it takes a lot of DB I/O, hence why most probjects don\'t use it

                                                          it just seems a bit odd that berkeley thought about most problems, except for basic ones that were even present with classic, i doubt a simple \"did you recieve WU xxxxx?\" just after the transfer while the scheduler has still got the request \"on it\'s mind\" as it were, wouldn\'t be that difficult if a full query would take too much in the way of resources

                                                          Profile Ananas
                                                          Send message
                                                          Joined: Jul 12 05
                                                          Posts: 222
                                                          Credit: 665,833
                                                          RAC: 0
                                                          Message 1839 - Posted 17 Feb 2006 16:06:03 UTC

                                                            Last modified: 17 Feb 2006 16:08:58 UTC

                                                            The TCP protocol with its twin byte counter (rcv/snd bytes) is the best example how a protocol can be made quite waterproof.

                                                            The problem that BOINC has might be that they seem not to have anything like a session ID, so if the client returned an ACK, the scheduler wouldn\'t know how to join his scheduler reply to the following client acknowledge.

                                                            The major difference to (for example) FTP is, that each client has its own forked instance and they communicate on a separate port so it\'s always clear where the stuff comes from and where it belongs to.

                                                            Lee Carre
                                                            Send message
                                                            Joined: Sep 30 05
                                                            Posts: 65
                                                            Credit: 7,033
                                                            RAC: 0
                                                            Message 1840 - Posted 17 Feb 2006 17:45:02 UTC - in response to Message 1839.

                                                              The TCP protocol with its twin byte counter (rcv/snd bytes) is the best example how a protocol can be made quite waterproof.

                                                              The problem that BOINC has might be that they seem not to have anything like a session ID, so if the client returned an ACK, the scheduler wouldn\'t know how to join his scheduler reply to the following client acknowledge.

                                                              The major difference to (for example) FTP is, that each client has its own forked instance and they communicate on a separate port so it\'s always clear where the stuff comes from and where it belongs to.

                                                              well why not use a higher level protocol and specify a session_id using that?

                                                              Profile Ananas
                                                              Send message
                                                              Joined: Jul 12 05
                                                              Posts: 222
                                                              Credit: 665,833
                                                              RAC: 0
                                                              Message 1842 - Posted 17 Feb 2006 19:43:33 UTC

                                                                Don\'t ask me - I always complain about conceptional problems but ... *sigh*

                                                                Lee Carre
                                                                Send message
                                                                Joined: Sep 30 05
                                                                Posts: 65
                                                                Credit: 7,033
                                                                RAC: 0
                                                                Message 1843 - Posted 17 Feb 2006 20:35:01 UTC - in response to Message 1842.

                                                                  Don\'t ask me - I always complain about conceptional problems but ... *sigh*

                                                                  lol, ah well, thanks for the info, maybe the devs don\'t think the gain is worth the effort (or DB I/O)
                                                                  anyway, we\'ll let adam have his blog back now :) sorry adam ;)

                                                                  Profile kadam
                                                                  Project administrator
                                                                  Avatar
                                                                  Send message
                                                                  Joined: May 25 05
                                                                  Posts: 589
                                                                  Credit: 38,614
                                                                  RAC: 0
                                                                  Message 1845 - Posted 17 Feb 2006 22:11:50 UTC - in response to Message 1843.

                                                                    Last modified: 17 Feb 2006 22:17:46 UTC

                                                                    anyway, we\'ll let adam have his blog back now :) sorry adam ;)


                                                                    Haha, Lee, not at all, this is an open blog! All is warmly welcome here! I think it would be good to wrap the communications, that belongs together in a something like a session. Hope they have it on their wishlist in Berkeley.
                                                                    ____________
                                                                    If you like BOINC, you may also find CaretCursor to be appealing.

                                                                    Profile kadam
                                                                    Project administrator
                                                                    Avatar
                                                                    Send message
                                                                    Joined: May 25 05
                                                                    Posts: 589
                                                                    Credit: 38,614
                                                                    RAC: 0
                                                                    Message 1847 - Posted 17 Feb 2006 22:21:16 UTC - in response to Message 1845.

                                                                      Listen up, everyone! I\'ve decided to reward Nightbird\'s efforts made on our forums. So, I\'d like to announce that from now on Nightbird is an official moderator on the forums. Give him a loud applause!!!
                                                                      ____________
                                                                      If you like BOINC, you may also find CaretCursor to be appealing.

                                                                      Profile Nightbird
                                                                      Forum moderator
                                                                      Avatar
                                                                      Send message
                                                                      Joined: Jul 12 05
                                                                      Posts: 920
                                                                      Credit: 114,924
                                                                      RAC: 0
                                                                      Message 1851 - Posted 17 Feb 2006 22:45:33 UTC

                                                                        Thanks, Adam.
                                                                        ____________

                                                                        Trulayne
                                                                        Avatar
                                                                        Send message
                                                                        Joined: Oct 16 05
                                                                        Posts: 36
                                                                        Credit: 11,377
                                                                        RAC: 0
                                                                        Message 1852 - Posted 17 Feb 2006 23:43:56 UTC

                                                                          Congrats, Nightbird!
                                                                          It is a status well deserved and earned by all your help in this forum.

                                                                          ____________

                                                                          Lee Carre
                                                                          Send message
                                                                          Joined: Sep 30 05
                                                                          Posts: 65
                                                                          Credit: 7,033
                                                                          RAC: 0
                                                                          Message 1853 - Posted 18 Feb 2006 4:42:05 UTC - in response to Message 1847.

                                                                            Give him a loud applause!!!

                                                                            *applauds loudly*
                                                                            it\'s well deserved :)

                                                                            Profile Ananas
                                                                            Send message
                                                                            Joined: Jul 12 05
                                                                            Posts: 222
                                                                            Credit: 665,833
                                                                            RAC: 0
                                                                            Message 1875 - Posted 18 Feb 2006 19:38:17 UTC

                                                                              Linux/x86 1.02
                                                                              Windows/x86 1.01

                                                                              I wonder if that\'s the reason why everyone gets this message \"there was work but I won\'t give it to you - hehehe\"


                                                                              If all WUs are 1.02 WUs, windows will get only the expired stuff in doggybags

                                                                              Profile kadam
                                                                              Project administrator
                                                                              Avatar
                                                                              Send message
                                                                              Joined: May 25 05
                                                                              Posts: 589
                                                                              Credit: 38,614
                                                                              RAC: 0
                                                                              Message 1883 - Posted 19 Feb 2006 1:35:42 UTC - in response to Message 1875.

                                                                                Ananas, WUs aren\'t assigned to a specific version of the program, so this should not be the source of problem. Although, I haven\'t got a figure who/what is behind this...
                                                                                ____________
                                                                                If you like BOINC, you may also find CaretCursor to be appealing.

                                                                                Profile Ananas
                                                                                Send message
                                                                                Joined: Jul 12 05
                                                                                Posts: 222
                                                                                Credit: 665,833
                                                                                RAC: 0
                                                                                Message 1884 - Posted 19 Feb 2006 8:07:05 UTC

                                                                                  Last modified: 19 Feb 2006 8:29:24 UTC

                                                                                  The source say that there\'s only one flag responsible for that : homogenous redundancy.

                                                                                  The database attribute that decides about that would be workunit.hr_class.

                                                                                  This should contain OS and CPU type and is compared against the matching sched_request tags (substring search in <vendor> / <os_name>).

                                                                                  The values are OR\'ed (well, actually added, should be OR\'ed though) from one from each group of those :

                                                                                  const int unspec = 0;
                                                                                  const int nocpu = 1;
                                                                                  const int Intel = 2;
                                                                                  const int AMD = 3;
                                                                                  const int Macintosh = 4;

                                                                                  const int noos = 128;
                                                                                  const int Linux = 256;
                                                                                  const int Windows = 384;
                                                                                  const int Darwin = 512;
                                                                                  const int SunOS = 640;

                                                                                  I get a lot of rejects on boxes having 387 - AMD+Windows, not 80387 ;-)

                                                                                  Profile kadam
                                                                                  Project administrator
                                                                                  Avatar
                                                                                  Send message
                                                                                  Joined: May 25 05
                                                                                  Posts: 589
                                                                                  Credit: 38,614
                                                                                  RAC: 0
                                                                                  Message 1889 - Posted 19 Feb 2006 13:08:17 UTC - in response to Message 1884.

                                                                                    Ananas, we don\'t have homogeneous redundancy. We had, but it caused a lot of problems, so we\'ve decided to turn it off and live with the side effect of various platform returning various results...
                                                                                    ____________
                                                                                    If you like BOINC, you may also find CaretCursor to be appealing.

                                                                                    Profile KSMarksPsych
                                                                                    Avatar
                                                                                    Send message
                                                                                    Joined: Dec 16 05
                                                                                    Posts: 115
                                                                                    Credit: 2,812
                                                                                    RAC: 0
                                                                                    Message 1891 - Posted 19 Feb 2006 13:59:01 UTC - in response to Message 1889.

                                                                                      Ananas, we don\'t have homogeneous redundancy. We had, but it caused a lot of problems, so we\'ve decided to turn it off and live with the side effect of various platform returning various results...


                                                                                      So if there isn\'t any HR, then why do I get the dreaded

                                                                                      2/19/2006 10:54:29 PM|SZTAKI Desktop Grid|Message from server: No work sent
                                                                                      2/19/2006 10:54:29 PM|SZTAKI Desktop Grid|Message from server: (there was work but it was committed to other platforms)
                                                                                      2/19/2006 10:54:29 PM|SZTAKI Desktop Grid|No work from project


                                                                                      error messages?

                                                                                      Just curious and trying to learn...

                                                                                      Kathryn
                                                                                      ____________
                                                                                      Kathryn :o)
                                                                                      The BOINC FAQ Service
                                                                                      The Unofficial BOINC Wiki
                                                                                      The Trac System
                                                                                      More BOINC information than you can shake a stick of RAM at.

                                                                                      Profile Nightbird
                                                                                      Forum moderator
                                                                                      Avatar
                                                                                      Send message
                                                                                      Joined: Jul 12 05
                                                                                      Posts: 920
                                                                                      Credit: 114,924
                                                                                      RAC: 0
                                                                                      Message 1894 - Posted 19 Feb 2006 19:48:40 UTC

                                                                                        Last modified: 19 Feb 2006 22:14:51 UTC

                                                                                        Situation :
                                                                                        - Windows/x86 users have this message \"(there was work but it was committed to other platforms)\"
                                                                                        - Is the server giving wus to Linux/x86 users ??
                                                                                        (no idea with Mac users)
                                                                                        look this post : http://szdg.lpds.sztaki.hu/szdg/forum_thread.php?id=221#1896
                                                                                        - according the status page, 16470 wus are ready to send (now 15330 -> 14,654)
                                                                                        - Performance diagram : GFLOP/s are decreasing quickly
                                                                                        - homogeneous redundancy is not enabled (??)

                                                                                        According the users, the problem [(there was work but it was committed to other platforms)] would be with platforms Windows, Linux and Solaris.

                                                                                        ____________

                                                                                        Profile Ananas
                                                                                        Send message
                                                                                        Joined: Jul 12 05
                                                                                        Posts: 222
                                                                                        Credit: 665,833
                                                                                        RAC: 0
                                                                                        Message 1899 - Posted 19 Feb 2006 21:13:48 UTC

                                                                                          Last modified: 19 Feb 2006 21:46:26 UTC

                                                                                          The HR check is the only place where the scheduler would issue exactly this error message.

                                                                                          From what I can see, neither DB.workunit nor DB.app have changed for quite some time so it\'s quite mysterious how the scheduler can get the idea to send this message.


                                                                                          Do you see a message like \"can\'t get hr_class for\" followed by a WUID and an error code in the scheduler logfile?
                                                                                          ______________________

                                                                                          I might have found something :

                                                                                          struct WORK_REQ { } has changed in server_types.h so if not all .o files using this include have been rebuilt, that might cause such a problem.


                                                                                          bool insufficient_disk;
                                                                                          bool insufficient_mem;
                                                                                          bool insufficient_speed;
                                                                                          bool excessive_work_buf; <=== NEW
                                                                                          bool no_app_version;
                                                                                          bool homogeneous_redundancy_reject;
                                                                                          bool outdated_core;


                                                                                          this new flag seems to be ffe., I cannot find a place where it\'s set to TRUE
                                                                                          ______________________

                                                                                          HR is definitely active, how else could this be explained :

                                                                                          All SUN results but SUN isn\'t really the majority here (yet).

                                                                                          gintaras46
                                                                                          Send message
                                                                                          Joined: Feb 17 06
                                                                                          Posts: 3
                                                                                          Credit: 19,315
                                                                                          RAC: 0
                                                                                          Message 1900 - Posted 19 Feb 2006 22:13:33 UTC

                                                                                            Last modified: 19 Feb 2006 22:14:05 UTC

                                                                                            Hi,

                                                                                            I have two machines, both Linux and Windows. On both clients I see same message

                                                                                            Mon Feb 20 00:21:01 2006|SZTAKI Desktop Grid|Requesting 8640 seconds of new work
                                                                                            Mon Feb 20 00:21:07 2006|SZTAKI Desktop Grid|Scheduler request to http://szdg.lpds.sztaki.hu/szdg/cgi-bin/scheduler succeeded
                                                                                            Mon Feb 20 00:21:07 2006|SZTAKI Desktop Grid|Message from server: No work sent
                                                                                            Mon Feb 20 00:21:07 2006|SZTAKI Desktop Grid|Message from server: (there was work but it was committed to other platforms)

                                                                                            ____________

                                                                                            Lee Carre
                                                                                            Send message
                                                                                            Joined: Sep 30 05
                                                                                            Posts: 65
                                                                                            Credit: 7,033
                                                                                            RAC: 0
                                                                                            Message 1901 - Posted 19 Feb 2006 22:24:02 UTC - in response to Message 1900.

                                                                                              Hi,

                                                                                              I have two machines, both Linux and Windows. On both clients I see same message

                                                                                              Mon Feb 20 00:21:01 2006|SZTAKI Desktop Grid|Requesting 8640 seconds of new work
                                                                                              Mon Feb 20 00:21:07 2006|SZTAKI Desktop Grid|Scheduler request to http://szdg.lpds.sztaki.hu/szdg/cgi-bin/scheduler succeeded
                                                                                              Mon Feb 20 00:21:07 2006|SZTAKI Desktop Grid|Message from server: No work sent
                                                                                              Mon Feb 20 00:21:07 2006|SZTAKI Desktop Grid|Message from server: (there was work but it was committed to other platforms)

                                                                                              i see the same on all 3 of my hosts (windows machines), don\'t think i\'ve had SZTAKI work for at least a few days

                                                                                              Profile kadam
                                                                                              Project administrator
                                                                                              Avatar
                                                                                              Send message
                                                                                              Joined: May 25 05
                                                                                              Posts: 589
                                                                                              Credit: 38,614
                                                                                              RAC: 0
                                                                                              Message 1911 - Posted 20 Feb 2006 14:39:30 UTC - in response to Message 1901.

                                                                                                Ok, some serious paranormal acitivity is going around here. I\'ve no idea what has happened. As I\'ve told you we don\'t have Homogeneous Redundancy turned on. Yet, when I checked today the database was full with WUs committed with HR turned on. I\'ve had them corrected, but I really have no idea how the hell they creaped in the database...
                                                                                                ...thanks for noting the problem, and sorry...
                                                                                                ____________
                                                                                                If you like BOINC, you may also find CaretCursor to be appealing.

                                                                                                Profile Ananas
                                                                                                Send message
                                                                                                Joined: Jul 12 05
                                                                                                Posts: 222
                                                                                                Credit: 665,833
                                                                                                RAC: 0
                                                                                                Message 1917 - Posted 21 Feb 2006 1:35:52 UTC

                                                                                                  Last modified: 21 Feb 2006 1:59:14 UTC

                                                                                                  Maybe get Mully and Sculder, the HR message is back.
                                                                                                  ________________________________

                                                                                                  Let\'s say my assumption is correct and it is somehow caused by the change in server_types.h.

                                                                                                  The mapping would be this then :

                                                                                                  bool insufficient_disk; <= insufficient_disk;
                                                                                                  bool insufficient_mem; <= insufficient_mem;
                                                                                                  bool insufficient_speed; <= insufficient_speed;
                                                                                                  bool excessive_work_buf; <= no_app_version;
                                                                                                  bool no_app_version; <= homogeneous_redundancy_reject;
                                                                                                  bool homogeneous_redundancy_reject; <= outdated_core;
                                                                                                  bool outdated_core; <= void

                                                                                                  So if the scheduler finds an outdated core like 4.43, 4.19 or so, it would set the flag \"outdated_core\" which ends up in \"homogeneous_redundancy_reject\" for the module that doesn\'t know that \"excessive_work_buf\" new has been inserted there.

                                                                                                  It\'s just an idea, my experience tells me that it has to be something like that rather than paranormal effects. (I\'m not familiar with Hungarian ghosts though - I know one Andras from there but he explained that he isn\'t a ghost)

                                                                                                  Profile Nightbird
                                                                                                  Forum moderator
                                                                                                  Avatar
                                                                                                  Send message
                                                                                                  Joined: Jul 12 05
                                                                                                  Posts: 920
                                                                                                  Credit: 114,924
                                                                                                  RAC: 0
                                                                                                  Message 1918 - Posted 21 Feb 2006 7:38:25 UTC

                                                                                                    Last modified: 21 Feb 2006 7:41:52 UTC

                                                                                                    No paranormal effects, no ghosts, no XFiles of course.
                                                                                                    Something is not working correctly and needs to be corrected quickly.
                                                                                                    With more and more registred members, we need wus and that wus flow smoothly.

                                                                                                    ____________

                                                                                                    Profile kadam
                                                                                                    Project administrator
                                                                                                    Avatar
                                                                                                    Send message
                                                                                                    Joined: May 25 05
                                                                                                    Posts: 589
                                                                                                    Credit: 38,614
                                                                                                    RAC: 0
                                                                                                    Message 1919 - Posted 21 Feb 2006 10:03:59 UTC - in response to Message 1918.

                                                                                                      I think it\'s not in connection with outdated clients, cause somehow the newly generated WUs have their hr_class (Homogeneous Redundancy) flag set, even though they shouldn\'t do so. That\'s why they don\'t get downloaded...
                                                                                                      ____________
                                                                                                      If you like BOINC, you may also find CaretCursor to be appealing.

                                                                                                      Post to thread

                                                                                                      Message boards : SZTAKI Desktop Grid : Ad@m\'s BLOG (page 3)


                                                                                                      Home | My Account | Message Boards


                                                                                                      Copyright © 2017 SZTAKI Desktop Grid