Computing Information Resources

• General Info

  • The Workbook is a comprehensive document that takes a user from getting an account to running a sample Beta job and beyond. The sample code uses the up-to-date release analysis-13b.

  • There is also a Frequently Asked Question (FAQ) page. It has BaBar-specifc topics such as how long is a release kept on disk to more general topics like dealing with Postscript files.

  • David Hitlin sent out a message to the entire collaboration on 8/13/99, in which he said:

    "We are immediately going to place controls on the number of
    jobs simultaneously accessing the database. This is a test which will,
    we believe, result in a substantial reduction in the wall clock time
    consumed by your jobs."

    This policy has been implemented. All jobs that access the Objectivity production federations must be submitted to the bfobjy batch queue. Interactive debugging is necessarily exempted.

  • The Augmented Tech Board decision on a batch priority scheme has been implemented.

• SCCS Related Items -- Since BaBar and SCCS work closely together, the dividing line is not always clear. You will find both generic information provided by SCCS and specific details provided by BaBar on some topics, e.g. the batch system.

  • You can call 650-926-COMP (650-926-2667) for a recorded status message.

  • Trouble calls should be made to the Help Desk at 650-926-HELP (650-926-4357) or by e-mail to Help Desk. Urgent problems reported by phone should get a response (but not necessarily a resolution) within 15 minutes.

  • Outages are posted to the SLAC Computing Outages and Changes HN. Non-critical scheduled outages are announced at least 24 hours in advance. It is often difficult to disseminate information on unexpected outages, esepcially to people off-site. If the above methods do not work, you might wish to check with IR-2. Please do this sparingly; shifters have a lot of work to do.

  • The intended use of the various pools of public Unix machines can be found in an SCCS document. Not all of them are suitable for every BaBar application. BaBar specific guidelines are available in this FAQ.

  • SCCS has a short and a more detailed introduction to the batch system. BaBar specific guideline is in an FAQ. There are two specific points you should be aware of. First, use the bldrecoq for building. Second, use the bfobjy queue for analboot and sp1analboot jobs.

  • You can request more personal AFS space by willing out the form.

• Objectivity Database

  • The DB Group's documentation page has links to many DB subjects.

  • There are also Objectivity related items in the FAQ, e.g. deleting a Federation when the release has been removed.

  • We have instituted regularly scheduled DB outages for sweeps and general maintenance. The outages are currently Monday or Tuesday 09:00 to 16:00. Adil often reports on what has been achieved in postings to the BaBar Computing Environment HN, e.g. #757 for the weekend of 9/4/99.

  • Jobs accessing Objectivity production federations must use the bfobjy batch queue, except for interactive debugging.

  • Jobs accessing the Kanga eventstore must use the kanga queue

  • What runs have raw+reco information? Disk space constraints make it impossible to keep the raw and reco information on disk for every run. The plan is to keep certain selected runs on disk for extended periods. See the lists provided for physboot, and for simuboot. Requests for additional runs should be sent to BdbSOS HN. In addition, a tool for users to stage back individual runs as needed is being developed, but is not yet ready for general use.

  • There has been significant progress on DB and PR development. See HN article by David Quarrie, and his contributions to the Forum meeting on 9/7/99.

  • The database browser provides a BaBar specific view of the Objectivity databases. See the Guide, and Yemi's presentation at 8/10/99 Forum meeting for details.

• Is this a known problem? And is there a known solution?

  • See if your problem is due to an outage, scheduled or otherwise. See SLAC Outages and Changes HN, and check the recorded status at 650-926-COMP (650-926-2667). The correlation between outage X and problem Y is often not obvious, even to the expert. It's something we have to learn.

  • BaBar's own computing outages are announced in Computing Environment HN.

  • Check that your account is properly set up at this time. Something could have changed since you last logged in. See the Workbook section on Setting Up your Account. Verify that the "chkuser" command gives the expected output.

  • Check that you are following the instructions for whatever you are trying to do. I have certainly failed to do so many times.

  • See if the problem you are encountering has been reported or perhaps even resolved. The BaBar Search Tool can run on the entire BaBar Web or on certain subsections. The HN Search Form operates on the whole HN tree or a specific forum. The Remedy problem tracking page has a very useful "Display Problems Based On Remedy Field values" button. It's near the bottomn. Also check the FAQ.

  • Some problems, and their solutions, have been posted to more focussed HyperNews groups, e.g. OO Databases, Physics Analysis, etc. You have to search.

• Problem Reporting -- There are many ways to report problems. It is not always easy to do, but picking the right one will help get them resolved faster. In all cases, it is important to include the actual symptoms in addition to any diagnoses that you might have made. There have been many cases where only the (incorrect) conclusions were reported, making it harder and slower to deal with the real problem.

  • The Prelim Bugs HN is a popular forum for reporting problems and frustrations. The good thing is that many people seem to pay attention and provide feedback. The bad thing is that the response may not be timely or even correct. If you choose to post your problems here, you are encouraged to post the release-specific ones as follow-ups to "bug anchors". This complements the HN Search Form and the page search provided by most Web browsers when looking for specific issues.

  • You may get better feedback by posting specific problems to some specialized HyperNews groups, e.g. Vertexing Tools, Simulation Production, etc.

  • Real confirmed problems should be entered into Remedy to faciliate problem tracking.

  • If you experience unusually sluggish response, it is very helpful to capture current status information. The "slow" command will do this.

  • For SLAC infrastructure problems, the best course of action is to report them to the appropriate email notification list. Most of the time this will be unix-admin@slac.stanford.edu A list of all SLAC computing-related problem report mailing lists is available at email technical support. If the problem is urgent, and requires immediate attention, you can call the SCCS Help Desk at 650-926-HELP (650-926-4357) and select the correct voicemail option to report urgent problems. This will page the on-call person. Please be sure the problem isn't a BaBar-specific one before reporting it to SCCS (prelimBugs is good for determining that if you are not sure). Additional information that may be helpful can be found on the SLAC unix pages.

  • For Desktop system problems (NT troubles or hardware problems on Dell computers), goto the BaBar Desktop Trouble Report page and follow the instructions there.