Dear Geoffroy > About your question OS/ version of OSCAR, there is no clear answer: it all > depends which OSCAR packages you select (as i said before). If you want to > make sure you can deploy the compute nodes, just select the core OSCAR > packages and try to deploy your cluster. So far, i had no bug reports about > the core itself, only a few configuration files were off.
I think decent solution would be installing everything except Torque and other failed packages and then adding them manually to the system. But since I have little more time to play with it until cluster deployment deadline I will follow along with your bugfixes and will try to react on them as soon as I can. I will benefit from it because I won't need to manually install anything and you along with HPC community will get tested CentOS-OSCAR pair. I'm looking forward to your reply. All regards, Nikita ----- "Nikita Andreev" <n...@kemsu.ru> a écrit : > > Can you log in to the nodes? > > Since step 3 fails I cannot proceed to nodes installation. So I have > no nodes installed and hence I cannot login to any of them. > > Could you advise me at this point some prior CentOS (or any other OS) > + OSCAR versions pair which I can possibly try? > > Best of luck, > Nikita > > -----Original Message----- > From: geoffroy.val...@free.fr [mailto:geoffroy.val...@free.fr] > Sent: Friday, February 11, 2011 8:55 PM > To: Nikita Andreev > Cc: oscar-users@lists.sourceforge.net > Subject: Re: [Oscar-users] Problem with GUI on CentOS 5.5 > > Dear Nikita, > > A configuration script from Torque (package i am currently trying to > fix) failed, i will have a look as soon as possible. Otherwise, please > do not pay attention to the messages "Error: No matching Packages to > list" > > http://svn.oscar.openclustergroup.org/trac/osca/wiki/faq_install#DuringStep4:BuildOSCARClientImagethefollowingmessageappearsmanytimes:Error:NomatchingPackagestolist. > > Can you log in to the nodes? > > Regards, > > ----- "Nikita Andreev" <n...@kemsu.ru> a écrit : > > > Dear Geoffroy, > > > > I've tried to unselect LAM and another error has showed up. Find > logs > > attached. GUI says that installation was unsuccessful. Console says > > that installation is complete but I clearly see errors above in > > console log. > > > > Regards, > > Nikita > > > > -----Original Message----- > > From: geoffroy.val...@free.fr [mailto:geoffroy.val...@free.fr] > > Sent: Thursday, February 10, 2011 10:41 PM > > To: Nikita Andreev > > Cc: oscar-users@lists.sourceforge.net > > Subject: Re: [Oscar-users] Problem with GUI on CentOS 5.5 > > > > Dear Nikita, > > > > Can you please unselect LAM and try to deploy the cluster? I opened > a > > ticket about LAM, i will have a look as soon as possible (i do not > > know why there is a dependency to libtorque, it does not make much > > sense to me). > > > > http://svn.oscar.openclustergroup.org/trac/oscar/ticket/600 > > > > Regards, > > > > ----- "Nikita Andreev" <n...@kemsu.ru> a écrit : > > > > > Dear Geoffroy, > > > > > > I've reinstalled OS and chosen unstable repo. Another dependency > > > problem has shown up. Find logs attached. > > > > > > Switcher still doesn't see MPI implementations. Don't know if it > > has > > > something to do with this dependency problem. > > > > > > Glad to be of help. > > > > > > Regards, > > > Nikita > > > > > > -----Original Message----- > > > From: Nikita Andreev [mailto:n...@kemsu.ru] > > > Sent: Thursday, February 10, 2011 11:55 AM > > > To: geoffroy.val...@free.fr > > > Cc: oscar-users@lists.sourceforge.net > > > Subject: Re: [Oscar-users] Problem with GUI on CentOS 5.5 > > > > > > > To really fix step 3, a modification of a script was needed. If > > you > > > are using Torque, i even advice you to switch to "unstable". > > > > > > > To really fix step 4, the scripts must be fixed so that > > everything > > > can run smoothly. Otherwise, OSCAR detects that a script failed > and > > > therefore cannot complete the step. Again, if you accept of using > > the > > > updated packages, switcher will not catch errors anymore and > > > everything should run smoothly. > > > > > > Image creation just failed with error Error: Missing Dependency: > > > torque-oscar = 2.1.10-4 is needed by package > > > torque-oscar-modulefile-2.1.10-4.noarch (unstable_rhel-5-x86_64) > > > > > > I will try to test unstable ASAP. > > > > > > > Also, when you question the competences of someone, this could > be > > > interpreted as insulting, there is nothing surprising there. So > > before > > > to question someone's expertise, the minimum you could do is to > > check > > > the position of the person in the project. > > > > > > Again, my apologies. I really didn't mean it. > > > > > > Regards, > > > > > > ----- "Nikita Andreev" <n...@kemsu.ru> a écrit : > > > > > > > Dear Geoffroy, > > > > > > > > I do not know where I've insulted you, but anyway if I've said > > > > anything inappropriate I would like to apologize. I greatly > > > appreciate > > > > your work as an OSCAR developer and I didn't mean to offend you > > in > > > any > > > > way. You're doing a great job. > > > > > > > > To my question. It's seems that there is some misunderstanding. > > > > Regarding problem with Torque in step 3 as I said in first > letter > > I > > > > solved it just by rebuilding opkg-torque-server package. I > > > downloaded > > > > it and changed .spec file so that it depended on > > > > torque-oscar-modulefile instead of torque-modulefile. And step > 3 > > > went > > > > fine from that point. > > > > > > > > Regarding step 4. I had GUI issue with error message in console > I > > > > quoted below. > > > > > > > > I've just updated OSCAR from 6.0.3 to 6.1.0 a minute ago by > `yum > > > > update` command and now GUI works fine. I'm able to run step 4 > > and > > > > waiting right now for image to be created. > > > > > > > > The only issue I still have is switcher. When I go into > > configuring > > > > switcher window from GUI I see: “No pkg_config were supplied by > > any > > > > OSCAR packages – nothing to configure”. I choose openmpi, mpich > > and > > > > lam to install so it has to see those mpi implementations. > > > > > > > > Regards, > > > > Nikita > > > > > > > > -----Original Message----- > > > > From: geoffroy.val...@free.fr [mailto:geoffroy.val...@free.fr] > > > > Sent: Thursday, February 10, 2011 10:15 AM > > > > To: Nikita Andreev > > > > Cc: oscar-users@lists.sourceforge.net > > > > Subject: Re: [Oscar-users] Problem with GUI on CentOS 5.5 > > > > > > > > Nikita, > > > > > > > > As i said, you may want to try to _not_ install the third-part > > > > packages that are creating the problems, the core of OSCAR > should > > > work > > > > pretty well (a user found a problem in C3 but we already have a > > fix > > > > that should be available very shortly). It is not a question of > > > > version of distribution but only that i cannot fix all the bugs > > > right > > > > away. > > > > > > > > So please, try to new packages. Obviously the problems you are > > > facing > > > > are because of these issues and if you are still blocked, it > > means > > > > that you did not really fixed them. I try to be reactive, to > help > > > you, > > > > so could you please try to do what i suggest to you? That will > > > really > > > > help. > > > > So to summarize, i pushed new RPMs for Torque in unstable, i > > really > > > > advice you to test them. I am about to push new packages for > > > Ganglia > > > > too, i really advice you to use these packages too. If you do > not > > > want > > > > to use these packages, i will not be able to help you. > > > > > > > > BTW, if you fixed a problem by yourself, could you at least > > > explained > > > > what you did? Otherwise i do not see how OSCAR could get > better; > > > this > > > > is an open source project, if i do not get feedback, there is > no > > > point > > > > of continuing on my side, i could maintain OSCAR on my side for > > my > > > own > > > > needs and do not distribute it. > > > > > > > > BTW, i am an OSCAR developer and the project chair, and i do > not > > > > appreciate the fact that you question my position in the > project, > > > this > > > > is almost insutling. My name is official displayed on the > > website, > > > and > > > > not difficult to find: > > > > http://svn.oscar.openclustergroup.org/trac/oscar/wiki/Team > > > > > > > > Regards, > > > > > > > > ----- "Nikita Andreev" <n...@kemsu.ru> a écrit : > > > > > > > > > > Based on your logs, many packages failed to install: > > > > > disable-services, ganglia, lam, maui, mpich, ntpconfig, > > openmpi, > > > > > opium, torque > > > > > > > > > > Please look below. I fixed torque problem by myself and > > > everything > > > > > went fine until step 4. disable-services, ganglia, lam, maui, > > > > mpich, > > > > > ntpconfig, openmpi, opium and torque installs just fine. > > > > > > > > > > > These packages are third-party packages and based on the > > logs, > > > i > > > > > cannot tell you why they were not installed successfully. So > > the > > > > first > > > > > thing i advice you to do is to unselect these packages using > > > > selector > > > > > and try to deploy again the cluster. > > > > > > > > > > Proceed through the logs down below and you will find that > > actual > > > > > error is some GUI bug in step 4. It pops up just after I push > > the > > > > > button in GUI. > > > > > > > > > > > Finally, about your question about PBS, remember that this > is > > a > > > > free > > > > > software and that i am working on OSCAR on my free time > > > > > > > > > > I understand. > > > > > > > > > > So the point is that I can't install OSCAR on top of CentOS > > 5.5. > > > > Since > > > > > it's a clean vanilla installation it should be common for > > CentOS > > > > 5.5 > > > > > for everyone. I fully understand that there is no official > > > support > > > > > from OSCAR and I'm on my own. But OSCAR have to install at > > least > > > on > > > > > some versions of CentOS. Can you give my some clue? Maybe I > > > should > > > > try > > > > > CentOS 5.4? It's a tedious and time consuming process of > > > > reinstalling > > > > > everything that's why I would be much more optimistic if this > > > > advice > > > > > will come from OSCAR developers. > > > > > > > > > > Regards, > > > > > Nikita > > > > > > > > > > ----- "Nikita Andreev" <n...@kemsu.ru> a écrit : > > > > > > > > > > > Dear Geoffroy, > > > > > > > > > > > > I want to emphasize that main problem is not Torque. It > seems > > > > like > > > > > > it's a problem with GUI which doesn't allow me to install > > OSCAR > > > > at > > > > > > all. Neither with Torque nor without it. > > > > > > > > > > > > Which version of OSCAR prior to version 6.1.0 is stable? > > > > > > > > > > > > Are all CentOS versions supported or only specific > versions? > > > > > > > > > > > > By the way, why Torque won't be supported? Which PBS will > be > > > > > supported > > > > > > then? > > > > > > > > > > > > Thank you for your time. > > > > > > > > > > > > Sincerely, > > > > > > Nikita > > > > > > > > > > > > -----Original Message----- > > > > > > From: geoffroy.val...@free.fr > [mailto:geoffroy.val...@free.fr] > > > > > > > > Sent: Wednesday, February 09, 2011 1:56 AM > > > > > > To: oscar-users@lists.sourceforge.net > > > > > > Cc: n...@kemsu.ru > > > > > > Subject: Re: [Oscar-users] Problem with GUI on CentOS 5.5 > > > > > > > > > > > > Hi, > > > > > > > > > > > > I am having problems with Torque since a while > (dependencies > > > > issues > > > > > > and problems building the binary packages). Unfortunately, > i > > do > > > > not > > > > > > have the time to fix that specific issue at the moment, > since > > i > > > > am > > > > > > focusing on the release of oscar-6.1.0 (which will _not_ > > > support > > > > > > Torque at first). > > > > > > > > > > > > A ticket is open related to that problem: > > > > > > http://svn.oscar.openclustergroup.org/trac/oscar/ticket/572 > > > > > > > > > > > > Once OSCAR-6.1.0 will be released, i will focus on fixing > all > > > > known > > > > > > issues related to OPKGs, the goal of 6.1.0 being to provide > a > > > > > robust > > > > > > OSCAR core. > > > > > > > > > > > > Regards, > > > > > > > > > > > > ----- "Nikita Andreev" <n...@kemsu.ru> a écrit : > > > > > > > > > > > > > I’m trying to deploy a cluster from CentOS 5.5 x86_64 > with > > > > OSCAR > > > > > > > 6.0.5. It’s a fresh install. I had a problem with > > > > > torque-modulefile > > > > > > > dependency with the following error: > > > > > > > > > > > > > > > > > > > > > > > > > > > > opkg-torque-server-2.1.13-1.noarch from > > > unstable_rhel-5-x86_64 > > > > > has > > > > > > > depsolving problems > > > > > > > > > > > > > > --> Missing Dependency: torque-modulefile is needed by > > > package > > > > > > > opkg-torque-server-2.1.13-1.noarch > (unstable_rhel-5-x86_64) > > > > > > > > > > > > > > > > > > > > > > > > > > > > I’ve resolved it by manually downloading > opkg-torque-server > > > and > > > > > > > rebuilding with torque-oscar-modulefile dependency which > I > > > > > believe > > > > > > is > > > > > > > correct. > > > > > > > > > > > > > > > > > > > > > > > > > > > > At the moment I have two issues I can’t resolve by > myself: > > > > > > > > > > > > > > > > > > > > > > > > > > > > 1. When I go into configuring switcher from the GUI I > see: > > > “No > > > > > > > pkg_config were supplied by any OSCAR packages – nothing > to > > > > > > > configure”. There should be something to configure since > I > > > > choose > > > > > > > openmpi, mpich and lam to install. > > > > > > > > > > > > > > > > > > > > > > > > > > > > 2. I can’t invoke 4th step. I get the following error > (find > > > log > > > > > > > attached): > > > > > > > > > > > > > > > > > > > > > > > > > > > > Tk::Error: Can't set -options to `ARRAY(0xd34c710)' for > > > > > > > Tk::Optionmenu=HASH(0xd360660): No -label at > > > > > > > > > > > > > > > > > > > > > > > > > > > > /usr/lib64/perl5/vendor_perl/5.8.8/x86_64-linux-thread-multi/Tk/Widget.pm > > > > > > > line 256 > > > > > > > > > > > > > > > > > > > > > > > > > > > > If CentOS 5.5 isn’t supported then please provide me with > > > > > supported > > > > > > > version. Is it 5.4? > > > > > > > > > > > > > > > > > > > > > > > > > > > > Another question would be why all necessary packages like > > > > torque, > > > > > > > maui, etc .. are under “Experimental” package set? Is > OSCAR > > > > 6.0.5 > > > > > > an > > > > > > > unstable version and I should rollback to version 5? > > > > > > > > > > > > > > > > > > > > > > > > > > > > Regards, > > > > > > > > > > > > > > Nikita > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > ------------------------------------------------------------------------------ > > > > > > > The ultimate all-in-one performance toolkit: Intel(R) > > > Parallel > > > > > > Studio > > > > > > > XE: > > > > > > > Pinpoint memory and threading errors before they happen. > > > > > > > Find and fix more than 250 security defects in the > > > development > > > > > > cycle. > > > > > > > Locate bottlenecks in serial and parallel code that limit > > > > > > > performance. > > > > > > > http://p.sf.net/sfu/intel-dev2devfeb > > > > > > > _______________________________________________ > > > > > > > Oscar-users mailing list > > > > > > > Oscar-users@lists.sourceforge.net > > > > > > > https://lists.sourceforge.net/lists/listinfo/oscar-users > > > > > > > > > > > > > > > ------------------------------------------------------------------------------ > > > The ultimate all-in-one performance toolkit: Intel(R) Parallel > > Studio > > > XE: > > > Pinpoint memory and threading errors before they happen. > > > Find and fix more than 250 security defects in the development > > cycle. > > > Locate bottlenecks in serial and parallel code that limit > > > performance. > > > http://p.sf.net/sfu/intel-dev2devfeb > > > _______________________________________________ > > > Oscar-users mailing list > > > Oscar-users@lists.sourceforge.net > > > https://lists.sourceforge.net/lists/listinfo/oscar-users ------------------------------------------------------------------------------ The ultimate all-in-one performance toolkit: Intel(R) Parallel Studio XE: Pinpoint memory and threading errors before they happen. Find and fix more than 250 security defects in the development cycle. Locate bottlenecks in serial and parallel code that limit performance. http://p.sf.net/sfu/intel-dev2devfeb _______________________________________________ Oscar-users mailing list Oscar-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/oscar-users
