climateprediction.net home page

The world's largest climate forecasting experiment for the 21st century.

Problems after SAP merger into CPDN


Advanced search

Message boards : Number crunching : Problems after SAP merger into CPDN

AuthorMessage
ChinookFoehn
Send message
Joined: Aug 7 04
Posts: 83
Credit: 410,895
RAC: 0
Message 33785 - Posted 15 May 2008 20:22:39 UTC

    chinookfoehn

    CPDN ID: 526
    SAP ID: 13836

    There doesn\'t seem to be a proper forum for this problem so I chose this thread.

    Earlier this month, I was issued an SAP task as the original computer failed to report. Things were going fine until...

    I could no longer access the SAP site. Eventually I could, but I kept getting this message as I did earlier today:

    15/05/2008 08:03:46|CPDN Seasonal Attribution Project|Message from server: Project is temporarily shut down for maintenance

    Then, it seems, today, SAP was merged into CPDN and now I get these messages:

    15/05/2008 10:35:34|CPDN Seasonal Attribution Project|Fetching scheduler list
    15/05/2008 10:35:39|CPDN Seasonal Attribution Project|Master file download succeeded
    15/05/2008 10:35:45|CPDN Seasonal Attribution Project|Sending scheduler request: Requested by user. Requesting 0 seconds of work, reporting 0 completed tasks
    15/05/2008 10:35:50|climateprediction.net|Scheduler request succeeded: got 0 new tasks
    15/05/2008 10:35:50|climateprediction.net|You used the wrong URL for this project
    15/05/2008 10:35:50|climateprediction.net|The correct URL is http://climateprediction.net/
    15/05/2008 10:35:50|climateprediction.net|You seem to be attached to this project twice
    15/05/2008 10:35:50|climateprediction.net|We suggest that you detach projects named climateprediction.net,
    15/05/2008 10:35:50|climateprediction.net|then reattach to http://climateprediction.net/

    15/05/2008 10:35:50|climateprediction.net|Already attached to a project named climateprediction.net (possibly with wrong URL)
    15/05/2008 10:35:50|climateprediction.net|Consider detaching this project, then trying again
    15/05/2008 10:35:50|climateprediction.net|Message from server: Invalid or missing account key. Visit this project\'s web site to get an account key.


    Sure enough, my CPDN.SAP work unit is now a CPDN work unit. As I already have another calculating...

    Rather interesting and quite the pickle. Both units show as CPDN, I get the error, I can only find my original account on CPDN and no sign of my SAP account.

    I have no desire to delete all CPDN task on my 2 computers as well as the single SAP task though the lesser evil would be cancelling the SAP task as I have no idea where it will report to now - though it is calculating at this moment (8.39% completed).

    CPDN task = hadcm3iozn_cq5p_2000_80_155899670_1
    SAP task = hadam3h_n_166s5_003a_0_1

    CPDN task = hadam3h_n_018s4_005c_0_1 (2nd computer)

    I do not wish to end up with an new id as this, likely, would split my CPID and lead to all sort of problems - something similar which occurred to me at Hydrogen@Home when their database failed and an old back-up was used resulting in the same error for myself and others. Eventually, through the efforts of the administrator, the error was corrected, the second i.d. cancelled, and my account was working properly - with only the loss of a few credits as a result.

    Here, I\'d rather not lose the credits I have.

    Please advise if you can combine the SAP task into my CPDN account or if I should abort the task.

    Les Bayliss
    Forum moderator
    Send message
    Joined: Sep 5 04
    Posts: 5129
    Credit: 8,442,487
    RAC: 4,532
    Message 33786 - Posted 15 May 2008 21:29:12 UTC

      SAP\'s last models were issued in December last year. It\'s apparently been decided that people have had enough time to complete them, and it\'s been shut down.
      (The server was also in a remote location, and hard to work on when problems occured.)

      Similar model types, but for a new experiment have ben available from the main site for several months.

      Those people still slowly working on models from the old SAP project will have to complete them on the main cpdn site.
      Just because the \"deadline\" is a year in the future, it doesn\'t mean that the results aren\'t needed as soon as possible. For the original SAP project, the research papers have been written.

      If anything more definite becomes known, a post will be made in News and Announcements.


      ____________
      Backups: Here

      ChinookFoehn
      Send message
      Joined: Aug 7 04
      Posts: 83
      Credit: 410,895
      RAC: 0
      Message 33787 - Posted 15 May 2008 21:46:31 UTC - in response to Message 33786.

        SAP\'s last models were issued in December last year. It\'s apparently been decided that people have had enough time to complete them, and it\'s been shut down....


        Be that as it might have been, I still was issued the SAP work unit, from the SAP server, at the start of this month so the project was not entirely shut down. I was issued it after the original 1-year deadline for the task expired (there was no work done on the task in that year). I would suppose that the database might still be viewable somewhere in Oxford - assuming anyone has interest in looking it up (no point really).

        This still does not address the issue of what I am to do with it as it is calculating and reporting but to a no-longer-existent SAP account and to no account on CPDN.

        Therefore, is the work being done actually accomplishing anything if there is no place to report to?

        My question still stands... do I abort the unit or is it worth the time and effort to have the task linked to my account?

        -ChinookFöhn

        Les Bayliss
        Forum moderator
        Send message
        Joined: Sep 5 04
        Posts: 5129
        Credit: 8,442,487
        RAC: 4,532
        Message 33788 - Posted 15 May 2008 22:02:21 UTC

          Last modified: 15 May 2008 22:03:38 UTC

          Still no word from the project people.
          And apparently there are other unexpected side effects showing up.

          As to what to do, I\'d suggest setting to No new tasks on SAP, and then Suspending any WUs still being worked on until further notice.


          ... I still was issued the SAP work unit, from the SAP server, at the start of this month

          The server would have been re-issuing failed models, but the last News item on the front page (21 December 2007), said that no new work was being issued, and that new work would come from the main cpdn site.
          Which was a hint to people to move to the main site, rather than clinging to the old site hoping for more work, for whatever reason.

          I\'m sorry that you\'re having problems, but it\'s come as a shock to ALL of us.
          The problem right now, is that it\'s night time in the UK, so it will be 12 hours or so before the project people return to work.
          ____________
          Backups: Here

          Profile Milo Thurston
          Forum moderator
          Volunteer developer
          Send message
          Joined: Mar 2 06
          Posts: 253
          Credit: 363,646
          RAC: 0
          Message 33789 - Posted 15 May 2008 22:03:39 UTC

            Last modified: 15 May 2008 22:04:23 UTC

            What has happened is that modifications to the SAP server (which has been giving us trouble for some time) have caused a temporary failure of BOINC to connect properly. Unfortunately we are not able to connect to the server to fix this due to network problems in Oxford\'s Atmospheric Physics department.
            Once the problems in Physics are sorted out I will deal with the SAP server; the earliest I would expect this to happen would be about 10 hours from now, but not much longer than that. A fix should be quite quick once I can gain access to the server, but there may be some disruption to the SAP web pages after that as more changes are made.
            Meanwhile I can only apologise for the inconvenience and suggest suspending your SAP workunits for the moment as Les has mentioned.

            ChinookFoehn
            Send message
            Joined: Aug 7 04
            Posts: 83
            Credit: 410,895
            RAC: 0
            Message 33793 - Posted 15 May 2008 22:50:35 UTC - in response to Message 33789.

              Last modified: 15 May 2008 22:51:15 UTC

              ...Once the problems in Physics are sorted out I will deal with the SAP server; the earliest I would expect this to happen would be about 10 hours from now, but not much longer than that. A fix should be quite quick once I can gain access to the server, but there may be some disruption to the SAP web pages after that as more changes are made. ...

              Is the SAP site actually in existence? I only get redirected to CPDN when I attempt to go there.

              ...Meanwhile I can only apologise for the inconvenience and suggest suspending your SAP workunits for the moment as Les has mentioned.

              No need to apologise.

              With the reconfiguration/transmogrification of the SAP unit into a CPDN unit... strange things have happened. I can suspend the ex-SAP unit but can not do anything with the CPDN unit. What happens to the SAP now happens to the CPDN. What I attempt to do to the CPDN occurs to the SAP. So, by suspending the SAP, the CPDN also is suspended.

              I suspect both units are now invalid.

              No need to rush on this... it is rather interesting what has occurred.

              As for still being connected to SAP - I left it so because of the CPID problems I was having. As I never expected any more work units to be issued from SAP, I didn\'t bother removing it after the CPID issue was resolved once the Hydrogen@Home issue was corrected.

              -ChinookFöhn

              Profile mo.v
              Forum moderator
              Avatar
              Send message
              Joined: Sep 29 04
              Posts: 2354
              Credit: 6,490,951
              RAC: 2,443
              Message 33794 - Posted 15 May 2008 22:50:53 UTC

                Last modified: 15 May 2008 22:52:31 UTC

                There\'s now a news announcement (top of Number Crunching) about this problem with instructions about what to do & what not to do. As soon as I\'ve copied that to all the news threads on all the forums, I\'ll add more technical details from Milo to explain what\'s happening.

                Many thanks to Chinook and others who\'ve reported this problem. Chinook, I hope you don\'t mind that I\'ve quoted your messages in the news post!
                ____________
                Cpdn news
                5 CPDN READMEs

                Profile Milo Thurston
                Forum moderator
                Volunteer developer
                Send message
                Joined: Mar 2 06
                Posts: 253
                Credit: 363,646
                RAC: 0
                Message 33795 - Posted 15 May 2008 22:56:13 UTC - in response to Message 33793.


                  Is the SAP site actually in existence? I only get redirected to CPDN when I attempt to go there. ...


                  It is still there. There was meant to be a temporary re-direction whilst we worked on upgrading the software on the SAP server; this was not supposed to have affected BOINC as well but things have gone a bit wrong.
                  I expect to have it fixed early tomorrow, though (it is just before midnight here).

                  ChinookFoehn
                  Send message
                  Joined: Aug 7 04
                  Posts: 83
                  Credit: 410,895
                  RAC: 0
                  Message 33796 - Posted 15 May 2008 22:57:56 UTC - in response to Message 33794.

                    Last modified: 15 May 2008 23:42:53 UTC

                    ... Chinook, I hope you don\'t mind that I\'ve quoted your messages in the news post!

                    Gosh! Golly! Does this mean I am published?

                    No problem at all. Hope you find a fix but I do not see how you can easily convert the ex-SAP-now-CPDN unit back into just an SAP unit as it seems to have melded with the prior CPDN unit.

                    addendum: This only occurs under the projects tab suspend. It works properly under the tasks tab suspend. I have suspended both work units for now.

                    Profile mo.v
                    Forum moderator
                    Avatar
                    Send message
                    Joined: Sep 29 04
                    Posts: 2354
                    Credit: 6,490,951
                    RAC: 2,443
                    Message 33798 - Posted 16 May 2008 0:20:14 UTC

                      I didn\'t quote your name with the messages, so it\'s your messages that are famous, not you!
                      ____________
                      Cpdn news
                      5 CPDN READMEs

                      ChinookFoehn
                      Send message
                      Joined: Aug 7 04
                      Posts: 83
                      Credit: 410,895
                      RAC: 0
                      Message 33800 - Posted 16 May 2008 3:04:41 UTC - in response to Message 33798.

                        I didn\'t quote your name with the messages, so it\'s your messages that are famous, not you!

                        Ah, the user behind the fame. Not a bad place to be.

                        Les Bayliss
                        Forum moderator
                        Send message
                        Joined: Sep 5 04
                        Posts: 5129
                        Credit: 8,442,487
                        RAC: 4,532
                        Message 33801 - Posted 16 May 2008 4:54:47 UTC

                          Searching on your name with Google, you\'re mentioned on 8260 pages.

                          ChinookFoehn
                          Send message
                          Joined: Aug 7 04
                          Posts: 83
                          Credit: 410,895
                          RAC: 0
                          Message 33802 - Posted 16 May 2008 6:45:12 UTC - in response to Message 33801.

                            Searching on your name with Google, you\'re mentioned on 8260 pages.

                            How interesting. I\'ve never thought to search my alias before. However, what is my fame when compared to a Les Bayliss search result of 289,000 mentions? Only 2.858%.

                            Interesting, on Google fr/ca there are only 8080 references compared to the 8260 at Google de/us/uk/au/es/ru and as for ChinookFöhn, there are 66 references reduced to 19 when I go to any of page 2/3/4.

                            Anyway, who would be interested in references to BOINC account data, statistics, and old posts - especially when so many seem to a foreign language version of the original language (English/German) data/post? Not me/myself/I. (This should screw up some the the translations)

                            -ChinookFöhn

                            Profile Milo Thurston
                            Forum moderator
                            Volunteer developer
                            Send message
                            Joined: Mar 2 06
                            Posts: 253
                            Credit: 363,646
                            RAC: 0
                            Message 33803 - Posted 16 May 2008 8:30:33 UTC

                              SAP is now back.

                              Profile mo.v
                              Forum moderator
                              Avatar
                              Send message
                              Joined: Sep 29 04
                              Posts: 2354
                              Credit: 6,490,951
                              RAC: 2,443
                              Message 33805 - Posted 16 May 2008 9:20:47 UTC

                                That was quick work, Milo!
                                ____________
                                Cpdn news
                                5 CPDN READMEs

                                ChinookFoehn
                                Send message
                                Joined: Aug 7 04
                                Posts: 83
                                Credit: 410,895
                                RAC: 0
                                Message 33809 - Posted 16 May 2008 15:40:35 UTC

                                  Morning/Afternoon Everyone.

                                  Something for you to ponder.

                                  Both tasks remain suspended.

                                  If I do an update on the ex-SAP task, I get sent to climateprediction.net.
                                  If I do an update on the CPDN task, the ex-SAP task is highlighted and I am sent to climateprediction.net.

                                  If I highlight the ex-SAP task and click the web-site button, I am sent to attribution.cpdn.org.
                                  If I highlight the CPDN task and click the website button, I am also sent to attribution.cpdn.org.
                                  Yet, in both cases, the web-site button displays climateprediction.net as both tasks display climateprediction.net as their project.

                                  Hopefully I am the only user with this problem as I suspect there may only have been a few users (hopefully just myself) that were sent tasks from the SAP server in last few weeks. If you get any ideas on how to untangle this web, do let me know and I give it a try.

                                  Not quite Alice in Wonderland but definitely ChinookFöhn in Climate Prediction dot Web.

                                  -ChinookFöhn

                                  Profile mo.v
                                  Forum moderator
                                  Avatar
                                  Send message
                                  Joined: Sep 29 04
                                  Posts: 2354
                                  Credit: 6,490,951
                                  RAC: 2,443
                                  Message 33810 - Posted 16 May 2008 16:00:17 UTC

                                    Hmmmmm.....

                                    I tried to check on which models you\'re referring to but your computers are hidden on CPDN and I couldn\'t get to your SAP project account because the SAP server seems to be completely dysfunctional again. I wonder whether Milo\'s trying out more ideas before the weekend?

                                    HADAM models from SAP come from the SAP server and send their trickles and file uploads to that same server.

                                    HADAM models from CPDN also come from the SAP server and presumably also send their trickles and zip file uploads back to the same SAP server.

                                    I think we\'d better wait to see what happens and what Milo says.

                                    If I had a HADAM model, whether from SAP or CPDN, I\'d suspend it and crunch something else for the time being.
                                    ____________
                                    Cpdn news
                                    5 CPDN READMEs

                                    ChinookFoehn
                                    Send message
                                    Joined: Aug 7 04
                                    Posts: 83
                                    Credit: 410,895
                                    RAC: 0
                                    Message 33813 - Posted 16 May 2008 17:43:07 UTC

                                      Now that the SAP site is back...

                                      SAP task: ex-SAP
                                      CPDN task: original CPDN

                                      The CPDN task did upload a trickle and given that I am 7 hours behind Zulu, this was sent after the SAP task was transmogrified into a CPDN task. I assume that I could resume the original CPDN work unit and it will report correctly (BOINC Manager just will not update correctly). As for the ex-SAP task...

                                      Anyway, I\'ll await your learned opinions on the matter.

                                      -ChinookFöhn

                                      Les Bayliss
                                      Forum moderator
                                      Send message
                                      Joined: Sep 5 04
                                      Posts: 5129
                                      Credit: 8,442,487
                                      RAC: 4,532
                                      Message 33814 - Posted 16 May 2008 19:15:48 UTC

                                        My cpdn HADAM3 models have a zip upload url of oerc.ox.ac.uk
                                        This is one of the main cpdn site\'s uploaders (uploadoerc), and I would think that the trickles from these models would also go to the main \"cpdn\" site as well.

                                        My 3 HADAM3 models have been uploading OK.

                                        Profile mo.v
                                        Forum moderator
                                        Avatar
                                        Send message
                                        Joined: Sep 29 04
                                        Posts: 2354
                                        Credit: 6,490,951
                                        RAC: 2,443
                                        Message 33816 - Posted 16 May 2008 19:55:31 UTC

                                          Chinook, that CPDN task is a HADCM and it successfully sent a trickle to a CPDN server less than 24 hours ago.

                                          From what you say, Les, the SAP server problem should only have affected people with HADAM models from the SAP project.

                                          It looks as if the CPDN HADAM models are only created on the SAP server and after that are dealt with entirely by CPDN servers. This is fortunate as it means the problem will have affected fewer people, and fewer models will have been lost through people following Boinc manager\'s suggestions.
                                          ____________
                                          Cpdn news
                                          5 CPDN READMEs

                                          ChinookFoehn
                                          Send message
                                          Joined: Aug 7 04
                                          Posts: 83
                                          Credit: 410,895
                                          RAC: 0
                                          Message 33817 - Posted 16 May 2008 20:40:52 UTC - in response to Message 33816.

                                            Chinook, that CPDN task is a HADCM and it successfully sent a trickle to a CPDN server less than 24 hours ago...

                                            Yes, my problem problem is that the SAP task was transmogrified from reporting to SAP to CPDN which now gives me 2 CPDN tasks which seem to be, partially, interlinked.

                                            The CPDN (HADCM) seems to update properly. The ex-SAP (HADAM) seems not to.
                                            As I previously posted... updating either one results in only the ex-SAP (HADAM) issuing the update - to the CPDN site, not the SAP site from where it was issued. The ex-SAP (HADAM) task is not shown on my CPDN account, only on my SAP account.

                                            Bizarrely, today, both tasks only go to the SAP site whereas, yesterday, both were going to the CPDN site - even though only climateprediction.net shows on the web-site button for both tasks.

                                            If no-one has anything against it, I\'ll keep the ex-SAP (HADAM) suspended until someone figures out out to remove the cross-links or informs me I should abort it, and resume the CPDN (HADCM) task. At least the trickles are working, the issued credit is being reported to statistic sites, such as BOINCstats, even though no updated credit is showing up on this BOINC Manager any more (115871 here, 116182 BOINCstats). I have not checked on the other computer.

                                            Hopefully, before the CPDN task is completed, the cross-linkage will be resolved or the SAP unit can be aborted so I can issue a No NEW Tasks command (only applies to the ex-SAP at the moment) and then reset the project once the task is finished. I also will not upgrade the BOINC Manager until that point as I can well imagine that would screw thing up royally - presently using v5.10.45.

                                            If I read nothing to the contrary, I\'ll resume the CPDN (HADCM) task tomorrow (Saturday) or earlier if you concur.

                                            -ChinookFöhn

                                            Les Bayliss
                                            Forum moderator
                                            Send message
                                            Joined: Sep 5 04
                                            Posts: 5129
                                            Credit: 8,442,487
                                            RAC: 4,532
                                            Message 33818 - Posted 16 May 2008 20:53:41 UTC

                                              Milo has done all that he can for this week.
                                              The department housing the SAP server has shut down for the weekend and there\'s no access until Monday.

                                              You\'re not the only one that received a model in the last couple of months, and the \"top\" 4 crunchers have dozens of models on their machines, which possibly don\'t get looked at often, so they might not be aware that there\'s a problem.

                                              Best that I can say is: Good Luck.

                                              ChinookFoehn
                                              Send message
                                              Joined: Aug 7 04
                                              Posts: 83
                                              Credit: 410,895
                                              RAC: 0
                                              Message 33819 - Posted 16 May 2008 21:29:31 UTC - in response to Message 33818.

                                                Milo has done all that he can for this week... Good Luck.

                                                Thanks for the thought but luck, I believe, won\'t have much to do with it, just analytical reasoning.

                                                The ex-SAP unit isn\'t due until November so there are a few months of time to try to solve it. Don\'t know if it is worth it other than as an intellectual exercise and for the knowledge it would bring for if no solution becomes available next week, then aborting and re-issuing the units, likely, is a faster method of obtaining the results.

                                                As for me, I am content to hold the the ex-SAP task until the end of July at which time, if a solution is found, then crunch it and something like Milky Way tasks, \'til it completes, and then the CPDN task with Milky Way until it completes, and then reset both CPDN and SAP on this computer.

                                                Have a nice weekend.

                                                -ChinookFöhn

                                                Les Bayliss
                                                Forum moderator
                                                Send message
                                                Joined: Sep 5 04
                                                Posts: 5129
                                                Credit: 8,442,487
                                                RAC: 4,532
                                                Message 33820 - Posted 16 May 2008 22:35:07 UTC

                                                  The November \"due date\" is artifical; something HAS to be put into that field for the other projects, and this project uses one a long time into the future.
                                                  The REAL due date was last year, before the project person\'s thesis was written.
                                                  The models now slowly being completed will go into the collection with the others, for future researchers to use.

                                                  The only think that will happen if you go past November, will be a message saying that the model is overdue, and to consider aborting it. Which isn\'t necessary.

                                                  This applies to ALL climate models.

                                                  [B^S] sTrey
                                                  Avatar
                                                  Send message
                                                  Joined: Jan 9 05
                                                  Posts: 30
                                                  Credit: 420,933
                                                  RAC: 0
                                                  Message 33823 - Posted 17 May 2008 1:51:25 UTC

                                                    My cpdn model has been suspended since before SAP broke, due to cpdn server problems and non-cpdn circs. SAP is now suspended on all my hosts, but it\'s still named climateprediction.net so if I force an update I still get the you\'re-already-attached-to-cpdn-please-detach messages.

                                                    So I wasn\'t clear if the redirect has truly been fixed... Does it appear safe to let normal cpdn models continue to run and contact the server, or do I need to wait for SAP to recover and the duplicate identity to go away first?

                                                    ChinookFoehn
                                                    Send message
                                                    Joined: Aug 7 04
                                                    Posts: 83
                                                    Credit: 410,895
                                                    RAC: 0
                                                    Message 33826 - Posted 17 May 2008 2:50:57 UTC - in response to Message 33823.

                                                      ...So I wasn\'t clear if the redirect has truly been fixed...
                                                      Nope.
                                                      Does it appear safe to let normal cpdn models continue to run and contact the server, or do I need to wait for SAP to recover and the duplicate identity to go away first?

                                                      All I can recommend is that you do as I did and look at your CPDN task under your computer and see if it did a trickle since the attempt to merge SAP into CPDN. If it did/does, I would say yes.

                                                      If it hasn\'t, and it is a HADAM unit... I\'d recommend you wait \'til what is accomplished next week. Of course my advice could be totally in error.

                                                      As I\'ve read nothing to the contrary, I shall re-start my CPDN (HADCM) in the morning.

                                                      -ChinookFöhn

                                                      KAMasud
                                                      Send message
                                                      Joined: Oct 6 06
                                                      Posts: 33
                                                      Credit: 399,398
                                                      RAC: 0
                                                      Message 33828 - Posted 17 May 2008 9:19:55 UTC


                                                        1) I was issued with a WU from SAP server this month.
                                                        2) Both my projects WU\'s have been effected.
                                                        What i can do but i don\'t have a spare machine, is to split both the projects from the combined folder into two different folders (can be done) then check up, but, alas will have to wait a month. Stuck i suppose :\'( 5 WU\'s of SAP and 3 160 year models? what a shame, i watch them grow, like i watch my kids grow.
                                                        Regards
                                                        Masud.

                                                        Profile MikeMarsUK
                                                        Forum moderator
                                                        Avatar
                                                        Send message
                                                        Joined: Jan 13 06
                                                        Posts: 1498
                                                        Credit: 6,796,400
                                                        RAC: 5,127
                                                        Message 33829 - Posted 17 May 2008 12:55:54 UTC


                                                          Thyme Lawn is experimenting with small edits to the project account XML file and client_state.xml to put things back where they should be. It seems to be fixing the problem on my affected PC, in any case...

                                                          ____________
                                                          I'm a volunteer and my views are my own.
                                                          News and Announcements and FAQ

                                                          Profile Thyme Lawn
                                                          Forum moderator
                                                          Send message
                                                          Joined: Aug 5 04
                                                          Posts: 1212
                                                          Credit: 10,212,294
                                                          RAC: 684
                                                          Message 33830 - Posted 17 May 2008 13:41:55 UTC

                                                            Last modified: 18 May 2008 14:40:08 UTC

                                                            The redirect caused the project name and scheduler URL for SAP to be changed to those for the main CPDN project. After testing some ideas out with MikeMarsUK the following sequence will sort out the problem.

                                                            Edit: the instructions have been changed because they relied on forcing BOINC to do a master file fetch before sending another scheduler request. This didn\'t always happen, causing the request to go to the CPDN server instead of the SAP server and undo all of the changes.

                                                            The new instructions are here.
                                                            ____________
                                                            "The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer

                                                            KAMasud
                                                            Send message
                                                            Joined: Oct 6 06
                                                            Posts: 33
                                                            Credit: 399,398
                                                            RAC: 0
                                                            Message 33831 - Posted 17 May 2008 14:11:22 UTC


                                                              Thyme i did as you suggested it started of Ok but reverted back.

                                                              5/16/2008 7:03:22 PM|CPDN Seasonal Attribution Project|Sending scheduler request: Requested by user. Requesting 0 seconds of work, reporting 0 completed tasks
                                                              5/16/2008 7:03:27 PM|climateprediction.net|Scheduler request succeeded: got 0 new tasks
                                                              5/16/2008 7:03:27 PM|climateprediction.net|You used the wrong URL for this project
                                                              5/16/2008 7:03:27 PM|climateprediction.net|The correct URL is http://climateprediction.net/
                                                              5/16/2008 7:03:27 PM|climateprediction.net|You seem to be attached to this project twice
                                                              5/16/2008 7:03:27 PM|climateprediction.net|We suggest that you detach projects named climateprediction.net,
                                                              5/16/2008 7:03:27 PM|climateprediction.net|then reattach to http://climateprediction.net/
                                                              5/16/2008 7:03:27 PM|climateprediction.net|Already attached to a project named climateprediction.net (possibly with wrong URL)
                                                              5/16/2008 7:03:27 PM|climateprediction.net|Consider detaching this project, then trying again
                                                              5/16/2008 7:03:27 PM|climateprediction.net|Message from server: Invalid or missing account key. Visit this project\'s web site to get an account key.


                                                              Just to keep life simple, what if we consider these as crashed WU\'s and re-run from back up?
                                                              Regards
                                                              Masud.

                                                              ChinookFoehn
                                                              Send message
                                                              Joined: Aug 7 04
                                                              Posts: 83
                                                              Credit: 410,895
                                                              RAC: 0
                                                              Message 33833 - Posted 17 May 2008 16:06:46 UTC

                                                                The same occurred with me other than...

                                                                WARNING! Do not restart any tasks if you have SAP tasks merged into CPDN.

                                                                as my original CPDN (HADCM) started up, started looking for HADAM data and errored out.

                                                                Of course I can not update the task to rid BOINC Manager of it as only the ex-SAP [HADAM] task issues updates.

                                                                I do not think it is worth trying to find a fix as the data must be corrupted.

                                                                Whether there is any value in obtaining the knowledge of how to correct this error when it seems obvious that the data in the intermingled work units is highly suspect.

                                                                That my CPDN task issued a correct trickle, once, after the fiasco must have been an anomaly.

                                                                I too vote for aborting all affected units and issuing a re-set project both to CPDN and SAP.

                                                                A shame but if the data is important, then it seems to me that it should be re-issued.

                                                                -Chinookföhn

                                                                Profile mo.v
                                                                Forum moderator
                                                                Avatar
                                                                Send message
                                                                Joined: Sep 29 04
                                                                Posts: 2354
                                                                Credit: 6,490,951
                                                                RAC: 2,443
                                                                Message 33834 - Posted 17 May 2008 17:02:36 UTC

                                                                  Hi sTrey

                                                                  It would be safer also to keep your CPDN model(s) suspended for the time being and crunch something else instead. If I were in your position I\'d back up the contents of the BOINC folder now, or certainly before restarting any climate models. And I wouldn\'t restart any of them until Milo tells us what the situation is on Monday.


                                                                  ____________
                                                                  Cpdn news
                                                                  5 CPDN READMEs

                                                                  Dagorath
                                                                  Avatar
                                                                  Send message
                                                                  Joined: Mar 3 06
                                                                  Posts: 96
                                                                  Credit: 353,185
                                                                  RAC: 0
                                                                  Message 33835 - Posted 17 May 2008 18:52:32 UTC - in response to Message 33831.


                                                                    Thyme i did as you suggested it started of Ok but reverted back.


                                                                    I think any manual change to client_state.xml reverts back unless you delete client_state_previous.xml before restarting BOINC. Could that be the reason Thyme Lawn\'s procedure doesn\'t seem to work?

                                                                    I would wait for Thyme Lawn to verify my theory. I may be wrong.



                                                                    ____________

                                                                    Profile mo.v
                                                                    Forum moderator
                                                                    Avatar
                                                                    Send message
                                                                    Joined: Sep 29 04
                                                                    Posts: 2354
                                                                    Credit: 6,490,951
                                                                    RAC: 2,443
                                                                    Message 33836 - Posted 17 May 2008 19:11:43 UTC

                                                                      Last modified: 17 May 2008 19:12:37 UTC

                                                                      Last year when my BBC model crashed with a \'Max CPU time exceeded\' message and I increased the fpops_bound figure in the xml file to give it more time to complete, I watched to see whether the figure reverted later. It didn\'t revert, so the client_state_previous file didn\'t need to be edited.

                                                                      The procedure I used for editing the xml file is described here for Windows:

                                                                      http://www.climateprediction.net/board/viewtopic.php?t=7215

                                                                      The edit procedure has now been tested by several members (for that xml file of course) and it definitely works. I wonder whether members are omitting some step of the procedure and this is causing the edit to revert later? Or perhaps some edits revert and some don\'t.
                                                                      ____________
                                                                      Cpdn news
                                                                      5 CPDN READMEs

                                                                      Profile Thyme Lawn
                                                                      Forum moderator
                                                                      Send message
                                                                      Joined: Aug 5 04
                                                                      Posts: 1212
                                                                      Credit: 10,212,294
                                                                      RAC: 684
                                                                      Message 33837 - Posted 17 May 2008 20:26:57 UTC - in response to Message 33835.

                                                                        Last modified: 17 May 2008 20:27:44 UTC

                                                                        I think any manual change to client_state.xml reverts back unless you delete client_state_previous.xml before restarting BOINC. Could that be the reason Thyme Lawn\'s procedure doesn\'t seem to work?

                                                                        client_state_prev.xml only comes into play if client_state.xml is corrupt. The fix failed for KAMasud (and probably chinooffoehn) because BOINC didn\'t issue the expected master file fetch before doing the scheduler update.

                                                                        I\'ve asked them to try a modification to the fix and have posted a modified set of instructions here.

                                                                        When I have confirmation that the modified fix works I\'ll modify the instructions here and on the SAP forum.
                                                                        ____________
                                                                        "The ultimate test of a moral society is the kind of world that it leaves to its children." - Dietrich Bonhoeffer

                                                                        ChinookFoehn
                                                                        Send message
                                                                        Joined: Aug 7 04
                                                                        Posts: 83
                                                                        Credit: 410,895
                                                                        RAC: 0
                                                                        Message 33839 - Posted 18 May 2008 1:47:58 UTC - in response to Message 33837.

                                                                          ...
                                                                          I\'ve asked them to try a modification to the fix and have posted a modified set of instructions here.
                                                                          ...

                                                                          The instructions did work, CPDN updated and SAP is back where it is supposed to be. The only difference I had, was that there was no change in my account_attribution_.cpdn.org.xml file.

                                                                          Until I am informed otherwise, I am am leaving the SAP task suspended.
                                                                          Alas, the CPDN task errored out and was lost.

                                                                          A rather novel experience but had I my druthers...

                                                                          Thank you Thyme Lawn for the correction and knowledge.

                                                                          -ChinookFöhn

                                                                          KAMasud
                                                                          Send message
                                                                          Joined: Oct 6 06
                                                                          Posts: 33
                                                                          Credit: 399,398
                                                                          RAC: 0
                                                                          Message 33840 - Posted 18 May 2008 7:16:25 UTC


                                                                            Hi Thyme, again did as you have suggested but the project name in Boinc Manager is still pointing towards CPDN Main? i.e. climateprediction.net. Even though i have edoted the name in account folder as per advise.
                                                                            I backup twice a day, what if i replace Boinc folder with a clean backup folder. Should do the trick. I have not tried it as yet due to WU\'s from LHC and RS.

                                                                            Hello Dagorath/ Seinfeld, at last you found peace at some project. LoL. Wonder, what magic the Mods did on you :).
                                                                            To Mods. You all, are the real driving force behind these climate projects.
                                                                            Regards
                                                                            Masud.

                                                                            Profile MikeMarsUK
                                                                            Forum moderator
                                                                            Avatar
                                                                            Send message
                                                                            Joined: Jan 13 06
                                                                            Posts: 1498
                                                                            Credit: 6,796,400
                                                                            RAC: 5,127
                                                                            Message 33842 - Posted 18 May 2008 7:56:24 UTC - in response to Message 33840.


                                                                              Hi Thyme, again did as you have suggested but the project name in Boinc Manager is still pointing towards CPDN Main? i.e. climateprediction.net. Even though i have edoted the name in account folder as per advise.
                                                                              ...


                                                                              I think you need to edit the account name in both the project account file and also the client_state.xml file (in the attribution section). The same XML tag appears in both.


                                                                              ____________
                                                                              I'm a volunteer and my views are my own.
                                                                              News and Announcements and FAQ

                                                                              KAMasud
                                                                              Send message
                                                                              Joined: Oct 6 06
                                                                              Posts: 33
                                                                              Credit: 399,398
                                                                              RAC: 0
                                                                              Message 33843 - Posted 18 May 2008 9:11:07 UTC


                                                                                Mike, i did make the changes in client_state.xml. The change in name occurs and every thing is Ok! until Boinc contacts the server. The name changes back? I open client_state.xml, it has reverted back.
                                                                                It seems that somehow the climateprediction.net genuine folder is controlling the ex SAP folder. I suspend climate and it suspends the ex SAP project, while it has forgotten all about its own WU\'s. I have to suspend them individually.
                                                                                Maybe Boinc is recording it some where? point me towards it and i can copy those contents, if that helps? Still, have a week of WU\'s from other projects, so no hurry.
                                                                                Regards
                                                                                Masud.

                                                                                Profile mo.v
                                                                                Forum moderator
                                                                                Avatar
                                                                                Send message
                                                                                Joined: Sep 29 04
                                                                                Posts: 2354
                                                                                Credit: 6,490,951
                                                                                RAC: 2,443
                                                                                Message 33849 - Posted 18 May 2008 11:22:25 UTC

                                                                                  Thyme Lawn has modified his original instructions here:

                                                                                  http://www.climateprediction.net/board/viewtopic.php?p=76424#76424
                                                                                  ____________
                                                                                  Cpdn news
                                                                                  5 CPDN READMEs

                                                                                  Profile mo.v
                                                                                  Forum moderator
                                                                                  Avatar
                                                                                  Send message
                                                                                  Joined: Sep 29 04
                                                                                  Posts: 2354
                                                                                  Credit: 6,490,951
                                                                                  RAC: 2,443
                                                                                  Message 33864 - Posted 18 May 2008 21:12:35 UTC

                                                                                    Copied from News thread:

                                                                                    SAP

                                                                                    This afternoon Milo managed to get into the Physics department. He hasn\'t explained exactly how he got in.

                                                                                    He has cancelled the SAP server redirect to the CPDN server and upgraded the server software. The SAP server is now completely up and running, though Milo can\'t guarantee its future reliability.

                                                                                    It should now be possible to run SAP models again and allow them to send trickles and uploads to the server.
                                                                                    ____________
                                                                                    Cpdn news
                                                                                    5 CPDN READMEs

                                                                                    KAMasud
                                                                                    Send message
                                                                                    Joined: Oct 6 06
                                                                                    Posts: 33
                                                                                    Credit: 399,398
                                                                                    RAC: 0
                                                                                    Message 33865 - Posted 19 May 2008 4:24:37 UTC


                                                                                      Yes mo.v it came on line midnight, our time. How did he get to it after sunset(Your Time)? On top of it, it was a Sunday. Any way, well done Milo.
                                                                                      I have started to appreciate this hard working secret weapon you people have.
                                                                                      Regards
                                                                                      Masud.

                                                                                      KAMasud
                                                                                      Send message
                                                                                      Joined: Oct 6 06
                                                                                      Posts: 33
                                                                                      Credit: 399,398
                                                                                      RAC: 0
                                                                                      Message 33920 - Posted 26 May 2008 13:52:25 UTC

                                                                                        Hi, just to bring a close to this thread. I have recovered all my WU\'s from backup.
                                                                                        Now, when will the SAP server come back on line?
                                                                                        Regards
                                                                                        Masud.

                                                                                        Profile marj
                                                                                        Forum moderator
                                                                                        Send message
                                                                                        Joined: Sep 13 04
                                                                                        Posts: 161
                                                                                        Credit: 284,548
                                                                                        RAC: 0
                                                                                        Message 33921 - Posted 26 May 2008 14:23:52 UTC - in response to Message 33920.

                                                                                          Hi, just to bring a close to this thread. I have recovered all my WU\'s from backup.
                                                                                          Now, when will the SAP server come back on line?
                                                                                          Regards
                                                                                          Masud.

                                                                                          It will probably be Tuesday before its back up as its a bank holiday weekend here and they can\'t get access till tomorrow.
                                                                                          ____________
                                                                                          _________________________________

                                                                                          Post to thread

                                                                                          Message boards : Number crunching : Problems after SAP merger into CPDN




                                                                                          Copyright © 2002-2014 climateprediction.net