CERN Accelerating science

Answer   -  

The online help is there, but the keyword to look for is 'strikethrough'.

Answer   -  

The process was interrupted while creating the temporary file, either because you pressed the ESC key or because there was some hardware or software failure or other media problem.

The original file should be still intact and unless your computer or workstation failed, the version of the file containing your current changes is still open in memory. Save the file to an alternate drive. If your computer or workstation did fail, you can use the original file.
(Note that any changes made in the last revision will be lost.)

Answer   -  

The only Ethernet cards that are currently (Dec. 2000) supported by the Win2000 installation at CERN are:

 Intel EtherExpress PCI PRO10+, PRO100B
 3c589 3Com EtherLink III adapter
 3c90X 3Com EtherLink PCI adapter
 

Answer   -  

Computers need to be registered correctly in LANDB for Windows 2000 to work and deploy applications. In this case, the computer had not been registered as portable from the network database.

You can check your computer configuration at http://cern.ch/winservices: select then "Computer account status", which will display in red incorrect settings.

The PC has to be registered correctly and you will have to wait for the active directory synchronisation to complete.

Answer   -  

Start - settings - control panel and double-click on 'network'. The 'identification' frame contains the PC's name.

Note: it is good practice to stick a label with the PC name on top of the screen (for quick identification).

Answer   -  

• click on the file
• save it to disk
• right-click on this saved file
• choose "scan for viruses"
• wait a few seconds until you get a result

If no viruses found, you can open the file by double-clicking.

Answer   -  

This means that the document is form protected. You have to select:

• From inside Word: Tools - unprotect document
• Tools - Options - View,
  un-tick the check box "Field Code"

 "the SMS Virtual Keyboard driver is preventing
 the machine from entering standby mode. Please close 
 all applications and try again. If problem persists 
 you may need to update the driver".
 

Answer   -  

This usually means the user has installed some additional software. Un-installing it should solve the problem.

In one particular case, FrameMaker was installed along with Adobe type manager 4.0. By removing Adobe type manager the problem was solved.

Answer   -  

This is a known problem with AFS (November 2000) which has some difficulties to handle very large amount of files. This often results in make processes which are un-killable and the only solution is for the system manager to reboot the node where this process is running.

Answer   -  

'rm *' or 'rm *.*' will only delete the ones you can see with 'ls', i.e. not including the .xyz files. 'rm .*' will do it.

Answer   -  

Indeed with the VXCERN shutdown (Christmas 2000) those terminals are NOT supported anymore. The only solution to connect to CERN computing services is to change device (e.g. PC connected to the CERN Network).

Answer   -  

As from Tuesday, Feb.13 2001, the interfaces xuserinfo and - for the Group administrators - xuserreg and xspaceadm have been disabled.

New versions have been introduced on Tuesday, Feb.6, 2001 on the Linux based service LXPLUS; they will still be available on RSPLUS to ease the transition.

N.B. The interfaces on LXPLUS are nearly IDENTICAL to the ones currently in use: the only minor differences would come from a change in the version of the ORACLE tool used.

Access to Web version:
You can also use the Web version of the interfaces: to do this, you must first register yourself by means of the "Web Reg." button which is displayed on the first page of xuserinfo/xuserreg/xspaceadm. Once registered, you can use the URLs http://userinfo.cern.ch(resp. http://userreg.cern.ch and http://spaceadm.cern.ch).

Answer   -  

In most LSF clusters (e.g. in Atlas), batch queues have different scheduling strategies. The policy is that shorter jobs have higher priorities than others. In practice that means that longer jobs on a particular node might be suspended (then resumed) to have shorter requests get executed. This is automatically done by the LSF system itself. That's why from time to time you may see requests in the SSUSP state like right now:

 
    > bjobs -u all -m shiftatlas | grep atlas032b
    556674 jleveque RUN   1nh        atlas010@at atlas032b   job2       Nov 
    9 10:29
    556620 mommsen  RUN   1nd        atlas005@at atlas032b   *pile_23.6 Nov 
    9 09:14
    556624 mommsen  SSUSP 1nd        atlas005@at atlas032b   *tpile_3.6 Nov 
    9 09:16
 

Answer   -  

The command is bpeek

If you look at http://consult.cern.ch/writeups/plusguide/ there is a very informative chapter on LSF. http://consult.cern.ch/cnls/230/art_pbatch.html has an example and also links to where you can find all the information on lsf and the online manuals.

Answer   -  

In order to request a certain amount of memory for a batch job you should use the -R option for this special resource, e.g.:

   
    bsub -R "mem > 200" ...
 

requests for the resource of at least 200 MB of memory.

Please do request a resource and _NOT_ a host. To check, if there are hosts available for a requested resource, you can use the command:

    lsload -R "mem > 200 && public" lxbatch
 

which shows some machines in the cluster LXBATCH, which are available for public use and which have more than 200 MB of memory.

More information on the batch system can be found at: http://wwwinfo.cern.ch/pdp/lsf/index.html and in the guides specified there.

Answer   -  

A size limit is imposed on messages sent to mailing lists. You can get this limit when you search this mailing list with SIMBA.

SIMBA is the Web interface that allows you to manipulate your own list(s).

Generally messages sent to mailing lists should not contain attachments (they should contain just URLs pointing to the files on the web).

Answer   -  

Users with personal workstations need to use 'userinfo' on the web or 'xuserinfo' on Unix, and set the 'hosts' section. The mail team will make the necessary update to allow Mailserver access without password typing.
Note: For the central hosts the system managers update a list by hand and the mail team takes it into account.

If you see any problem with userinfo, i.e. if this doesn't result in a resolution by the next day, or if on a central server your are asked for a password, then contact the Helpdesk (helpdesk@cern.ch), mentioning the name of the new workstation added to the cluster which requires the typing of MailServer password every time you wish to use pine, or the cluster node. In that case we must ask the system managers to make the necessary manual entry in the table of cluster workstation members.

Answer   -  

The sender must changes the view/character set/ to Western AND set 'set default character set' to make this change effective.

Answer   -  

The main contact point for all kind of general Mail issues is the Computing Helpdesk that can be reached through mail.support@cern.ch (or Helpdesk@cern.ch).

Some more specific e-mail addresses can be used for specific subjects:

• Listbox questions (mailing lists): listbox.support@cern.ch
• Calendaring questions: service-calendar-pilot@listbox.cern.ch
• Roaming questions: service-roaming-pilot@listbox.cern.ch
• LDAP questions: proj-ldap@listbox.cern.ch

Answer   -  

There is not yet any official/recommended solution for "calendaring".

A "Netscape CS&T Calendar" project had started (described at http://cern.ch/proj-ldap/calendar.html) but unfortunately IT recommend to discontinue the use of the Netscape calendaring pilot as it has no future with the supplier.

Situation on January 2001:
An alternative is being proposed by IT/IS to the IT management and hopes are to have a new pilot ready in March. Considered solutions enable Unix/Windows interoperability, and one possible one is based on Microsoft Exchange server, with both Outlook and Web access.

Answer   -  

Clearly, if one person listens to the radio there is not too much traffic. If several people, and the people are all on a shared 10 megabit Ethernet then there can be an effect.

There is also a difference between listening to a particular broadcast and tuning into an all-day music station!

Video, of course, is much more traffic.

  dvips: ! Couldn't find header file ma1r8a.pfb
 

Answer   -  

Nothing has changed with the TeX installation for almost two years (1999). This error (which comes up more and more) is due to the fact that users want to include EPS files that were generated on Windows, where the Arial font sits in the system, and so is "freely" available. When going to Linux or Mac, this font is not present.

In most cases it is enough to generate an empty file for these font files, e.g., the command "touch ma1r8a.pfb" will create such a dummy file in the current directory. If you are lucky, and the font outlines are actually included as a resource in the EPS file, everything will be all right. If not, the picture will have to be regenerated with the font included, since the message from dvips tells you that the font in question is "somehow" referenced.

Conclusion: there is no general solution to that problem. If the font is included in the EPS file(s), you can also eliminate the comment specifying that font resource in the EPS file, so that dvips no longer complains, or else generate empty font files as indicated above. The fact that "it works" on another system might just mean that such the file in question is indeed available on that particular system (e.g. Windows).

Answer   -  

Documentation (manual) for Geant3 is available on paper (from the Computing Helpdesk), or from the Web page http://wwwinfo.cern.ch/asdoc/Welcome.html.

The Geant4 documentation is only available from the Web (not on paper) at URL:

 http://cern.ch/geant4/G4UsersDocuments/Overview/html/index.html
 

In this page you will find the individual documents in both browsable (HTML, PDF) and printable format (PostScript, PDF).

 g77 chf77.f -o chf77 `cernlib packlib mathlib`
 

  ld: warning: symbol `pawc_' has differing sizes:
          (file /var/tmp/ccSHE7Xf.o value=0x3a980; file
  /cern/pro/lib/libpacklib.a
  (hbook1.o) value=0x9c40);   
          /var/tmp/ccSHE7Xf.o definition taken
 etc...
  Undefined                       first referenced
   symbol                             in file
  __aintf                             /cern/pro/lib/libpacklib.a(dztyp.o)
  __s_stop                            /cern/pro/lib/libpacklib.a(zend.o)
 etc...
 

Answer   -  

There are two issues in your compilation procedure:

1. The warnings about changing the length of the labelled common. You probably should not do it, but the linker usually does the correct thing and takes the longer one.
2. You are linking with g77 and Cernlib (and consequently HBOOK) is built with Solaris f77, so you get unsatisfied externals for the f77 run-time library. You should really compile and link you application with f77. Otherwise, you must specifically include the f77 runtime libraries at the end of your link sequence.

A more general remark is that using the portable CERN "hepf77" command from ASIS (see http://consult.cern.ch/qa/1582) will get you the correct compiler (and options), used to build Cernlib, on all our systems at CERN.

Answer   -  

Problem licenses with CE application can happen when users unknowingly take licences for a long time - e.g. either because an xterm session may have crashed - or because they have iconised a window. Nevertheless, some licences will also be used for long times by a single user, specially for long calculations on PaRC.

A new tool can help users - specifically in the case you cannot obtain a licence and would like to know who has all the available licences. This tool is available under Unix and Windows.

Under Unix, it is called cestat: it will be in your path after you have typed 'parc_apps' (alternatively the full path is : /afs/cern.ch/project/parc/bin/cestat).

On Windows (95 and NT) the same tool is available under:

 
  Start menu -> Computer Aided Engineering -> 
  Other tools -> Application Statistics
 

It should be quite intuitive to use and there is a small on-line help.

First you should choose a product and click on it. Then a list of features will be displayed. Click on a feature and you will see a list of userids currently using that feature. By clicking on a userid you will see the information about the person concerned.

Please note: only a subset of applications are included in cestat (more will follow). Naturally only those applications for which there is a finite number of licences is included.

Answer   -  

You get all information from the Fluka web site, at:

 http://cern.ch/fluka/
 

Answer   -  

ASIS installs multiple versions of gcc and the "gcc -v" option selects the appropriate version.

 lxplus012 ~[142]> perl -e 'use DBI'
  
 Can't locate DBI.pm in @INC (@INC contains:
 /usr/lib/perl5/5.00503/i386-linux
 /usr/lib/perl5/5.00503
                    /usr/lib/perl5/site_perl/5.005/i386-linux
 /usr/lib/perl5/site_perl/5.005 .)
                    at -e line 1.
                    BEGIN failed--compilation aborted at -e line 1.
 

Answer   -  

This is because on LXPLUS your perl command comes from the system (/usr/bin/perl) and it does not have the DBI module. You should better use /usr/local/bin/perl instead.

Unfortunately /usr/bin/perl is needed by some system tools and it cannot be removed. /usr/local/bin/perl (from ASIS) is the PERL version (common on all architectures) to use in scripts. This is usually not a problem, since the scripts contain explicitly which PERL to use (in the first #! line). But when using PERL on the command line, it is necessary (when there is a system version like on Linux) to specify which PERL version to use. Especially if you need modules like DBI, you should take the ASIS version (modules such as DBI which are in ASIS and not in the system).