CERN Accelerating science

This website is no longer maintained. Its content may be obsolete. Please visit http://home.cern/ for current CERN information.

CERN home pageCERN home pageDocuments by ReferenceDocuments by ReferenceCNLsCNLsYear 2000Year 2000Help, Info about this page

Contents

Editorial Information
Editorial
If you need help
What our Users Have to Say ...
Announcements Physics Computing Desktop Computing Internet Services and Network Scientific Applications and Software Engineering The Learning Zone
  • Questions and Answers from the Computing Helpdesk
User Documentation
Previous:The Learning Zone
Next:User Documentation
 (See printing version)



Questions and Answers from the Computing Helpdesk

Collected by the IT/User Support Team

Nicole Crémel (Editor) , IT/User Support


Abstract

This is a collection of Questions & Answers that have been treated by the Computing Helpdesk managed by the User Support group.

N.B. The number in parentheses refers to its relative numbering in the User Support "Question-Answer" database, at URL:
http://cern.ch/consult/qa/NNN
where NNN is the problem identifier (number).


 


PC support  -  

(Hannes Schwarzbauer  IT/User Support)

Question (QA 284)   - Reset a profile on Windows NT

I am registered as a Windows NT administrator. What is the procedure to be followed to reset a profile (when it is corrupted)?

Answer   -   The procedure is the following:
  • logout
  • delete roaming profile
  • login and delete local profile
To delete roaming profile: access "NICE NT Domain Administration" (which had been added in the Domain tools list) in:
 
     Start - System Management - NICE NT Domain Administration 
 

The tool only deletes the server copy of the profile and administrators still need to delete manually the local copy. The recommended way to delete the local copy is to log in as a local administrator and use the utility in

 
     Control Panel - System - User Profiles 
 
select the user account, and press the delete button. This will delete all the registry entries associated to the profiles.
N.B. Deleting the files in \profiles\ is not enough and you run the risk of creating multiple profiles with the same name and extensions like .000, .001, etc.

Notes:

  1. In case of problems you may also call local support for your division.
  2. To reset your profile on Windows 95 see http://consult.cern.ch/qa/193

Question (QA 287)   - ftp from NICE

How can I ftp from NICE?

Answer   -   The standard way is to use:

Start - More applications - Communications - ftp

and choose the host you want. You may also use Exceed in order to ftp to a public unix service. This method is described in:

http://consult.cern.ch/qa/0022


Question (QA 290)   - 'recover unaccessible files' on NICE

I cannot access certain files/folders on NICE that I could previously. How can this be corrected?

Answer   -   Try using the 'recover unaccessible files' option in:
 
 Start - more Applications - system configuration -  
 recover unaccessible files 
 
Note that this is only possible under Windows 95.

Question (QA 319)   - Lock X session with Exceed

How can I lock my X session when using Exceed from NICE?

Answer   -   First you should note that locking your X session will only lock the UNIX (e.g. PLUS service) part, but your PC will not be locked (look at the end for more details).

Having said that, the first thing to do is to make sure you have enabled on NICE "Allow Clients to Modify Host Access Control List" with the menu

 
  Start / More Appl. / Exceed X Terminal v6... / Tools 
 
and then select the tools "Xconfig" and "Security".

You may then lock the screen on the UNIX side (i.e. any PLUS or WGS service) from the usual menu entry (System menu and entry "Lock Screen") when using the HEPiX/X11 environment.
N.B. Alternatively, you can also run by hand the UNIX command 'xlock -remote'.

As said at the beginning this will only lock the UNIX side. In order to lock the PC screen completely, one must either:

  • On NT: CTRL_ALT_DEL and click on 'Lock Workstation'
  • On W2000: CTRL_ALT_DEL and click on 'Lock Computer'
  • On W95: the lock facility does not exist and the only possible workaround is to use a password protected screen saver and let it click in after one minute (for instance); to set this Start - Settings - Control Panel and double click on 'Display', select 'Screen Saver', the picture you like, 'Password protected' and the 'Wait:' time you want.

Question (QA 330)   - Windows 2000 account

How do I get an account for Windows 2000 and get my machine registered to use the new system?

Answer   -   For the account and registration for a machine please see URL:
 
 
 http://cern.ch/winservices/docs/W2KSetup/setup.asp 
 
This explains all you need to know about your account name and password.

For more in depth information on the Win2000 project at CERN, see http://cern.ch/winservices/default.asp which may help you further.


Question (QA 347)   - Netware to Windows 2000 Migration

What is going on concerning the migration of Netware to Windows 2000 on the CERN servers?

Answer   -   The current situation is explained at the following URL which also gives detailed explanations, a few hints for users and a schedule of what will go on in the future:
 
 
 http://cern.ch/homepage/migration.htm 
 

 


UNIX Support  -  

(Nicole Crémel and Roger Woolnough  IT/User Support)

Question (QA 142)   - LINUX server installation

Is there some documentation about how to install a LINUX server with a CERN standard configuration ?

Answer   -   All should be explained in the "Linux at CERN" Web page, at URL:

http://wwwinfo.cern.ch/pdp/ose/linux/linux_CERN.html

Question (QA 286)   - Linking an application on ATLAS

I have problems (errors) when linking an application on ATLAS

Answer   -   First thing to check is that some environment variables used by the loader (command ld) are not wrongly defined (e.g. in your login script). Examples of variables to check are LD_LIBRARY_PATH and LDOPTS

Question (QA 288)   - Long queueing time for batch

I am experiencing an unusually long queueing time for submitted batch jobs. Normally they begin to execute within minutes, instead of remaining pending with messages like "job forwarded to a remote cluster" or "awaiting allocation replies" for a matter of anything from 2 to 6 or 7 hours.

Answer   -   This can be an operational problem (especially if many users report it) and the best approach is to alert Serco (via the Helpdesk). To know more about it (and check if this is really an operational problem):
  • identification: command 'bhosts rsbatch' on RSPLUS shows a RSV column which number add up to a total of more than a few units (e.g. 10 or 15).
  • solution: run 'badmin reconfig' on RSPLUS.

Question (QA 291)   - AFS "Cache partition is FULL - Decrease cachesize!!!"

With the "ls" command on some AFS directory I am getting the following weird message:

 
   *** Cache partition is FULL - Decrease cachesize!!! *** 
 

Answer   -   We have an old diagnostic from Transarc on this namely:
 
 "The problem is occurring because we copy the entire file into the 
 local cache when one person on a machine has the file open and  
 someone else on that same machine is deleting it.  When you see  
 this error message, you can confirm this by doing a "ls -la" in the  
 directory where the file was deleted and look at the size of the  
 .___afsXXXX files." 
 
It looks as it is a known "feature". When the application using the file stopped the problem should go away.

Question (QA 293)   - AFS or network problem ?

How to check if there is an AFS or network problem ? I am getting many "connection timed out" messages.

Answer   -   In order to help the AFS team (AFS.Support@cern.ch) for a proper diagnostic you can check and give them the following information:
  • command ping afsxx on the server (e.g. ping afs5 should give time numbers between 0 to 2 ms. Numbers in 10s or very erratic mean trouble. (N.B. Command fs whereis shows on which server your files are).
  • command traceroute afsxx shows the network hops and which one is slow.
  • command ping rsplus01 (or any other central machine) can be executed for comparison (and check if there is a general network problem).
  • command fs checkservers shows if there is an AFS server problem.

Question (QA 294)   - LSF command bpeek

Is there a possibility with LSF to look into files which are opened during the run (like it was possible with lltail or llmore on sp2pub)?

Answer   -   Yes, using the command bpeek.

Question (QA 300)   - bin directory on atlas nodes

In a machine such as the atlas09 machine I have used to create the archive test in the bin directory. However I cannot see this file is not seen in the same directory on atlas008. Why?

Answer   -   To ensure a correct separation of files for appropriate architectures, the AFS directories use links to architecture specific bin directories. This can be seen if you change directory to the bin directory and issue the following command:
 
 [lxplus001] ~ % cd bin 
 [lxplus001] ~/bin % /bin/pwd 
 /afs/cern.ch/user/t/tjs/.@sys/bin 
 
As you see, there is an @sys in the path. This is expanded up to a string related to the Operating System, which is hp_ux102 on atlas09 (for HP-UX) and i386_linux22 on atlas008 (for Linux). This is intentional, since a binary file made on an HP will not run on a Linux box! If you are dealing with scripts, not binaries, then you should use the 'scripts' sub-directory instead, as this is a real directory (and the same across different Operating Systems).

Question (QA 302)   - IT policy regarding Linux

What is IT policy regarding Linux ?

Answer   -   The Linux for the 2000s (L2K) Programme team wrote a "CERN/IT Linux Activities Survey". This document should list all the activities around Linux in the CERN IT division (as well as official statements from the IT management).

You can find this document on the Web at URL:

  
       
 http://cern.ch/linux/l2k/docs/ilas.html 
 

Question (QA 303)   - hsm command on a private LINUX desktop

I cannot run the hsm command on my PC (private LINUX desktop) and get the error message 'configuration file not found' for some of the hsm commands.

Answer   -   To fully use the hsm command on a private LINUX desktop you need to be using an AFS login, that has an associated DCE service account, and have a copy of a small system file named /etc/shift.conf.

Note: this has to be installed as root but can be copied from any LXPLUS node. A few times per year this file is changed on LXPLUS that may affect hsm, so users who have this file and still get errors should first take a fresh copy.


Question (QA 317)   - xuserinfo on Linux

If I execute xuserinfo on LXPLUS (Linux) I get something very primitive and quite different from what I can get on RSPLUS. Why?

Answer   -   This is because there are NO Oracle tools available on Linux yet. The problem will disappear if Oracle was to deliver a software long overdue.

To use the "real" tools, you must run xuserinfo/xuserreg on another PLUS service or NICE (with the button called "Web Reg." on the first page of the interface). After this has been done, userinfo/userreg can be invoked via Netscape at URL:

 
 http://userinfo.cern.ch 
 (resp. http://userreg.cern.ch). 
 

A request was made to remove the command from the ASIS product list on LINUX, until it works, and to avoid confusion.


Question (QA 320)   - Command bkill

I cannot kill some batch jobs with the command bkill on LXBATCH. Can you please kill these jobs for me?

Answer   -   Please note that if you cannot kill your own jobs with bkill then it means that nobody can kill the job either: the process is waiting in kernel state and the only thing that can be done about it is rebooting the system. Another explanation can be that the communication between the submission cluster and the execution cluster is interrupted, and the bkill command cannot be sent to the execution node. Normally the command lsclusters shows when there are such interruptions.

In any cases, the Helpdesk cannot kill the job for any user: the only think they can do is to pass the message (with as precise information as possible, and at least the execution node) to the system administrator, so that they are aware of the problem, and schedule eventually a reboot for this system.


Question (QA 322)   - Service interruption

Is there a web page where users are notified of possible interruption of the tape service, and of its estimated duration ?

Answer   -   You can find some information about the status of tape servers in the TVScreen Web page, at URL: http://tvscreen.cern.ch/ExpTvScreen.html, under 'The SHIFT and CORE services'.

Question (QA 329)   - Responsible for tools in /usr/local/bin

Can I see who is responsible for the tools / scripts program which are in /usr/local/bin (e.g. /usr/local/bin/cuserinfo) and whether it is a tool provided by ASIS or not?

Answer   -   You can see this with the command ASISInfo --belong <tool_name>.

E.g.:

 
 [rsplus08] ~ > ASISInfo --belong /usr/local/bin/cuserinfo 
 ERROR:File /usr/local/bin/cuserinfo in ASIS - No such file or directory 
 [rsplus08] ~ > ls -l /usr/local/bin/cuserinfo 
 -rwsr-xr-x   1 regist   c3         10331 Mar 25 1998  /usr/local/bin/cuserinfo 
 [rsplus08] ~ > phone regist -F 
 23932;            Ccdb DATABASE                 72995               IT      31   3-008;Ccdb.Database@cern.ch 
 
which means that this is managed by the Ccdb DATABASE service responsible.

For /usr/local/bin/gs you get:

 
 [rsplus08] ~ > ASISInfo --belong /usr/local/bin/gs 
 GNU.MISC/ghostscript-5.50 
 
which shows that it is a GNU product provided by the ASIS team.

 


Mail support  -  

(Michel Franceschi and Miguel Marquina  IT/User Support)

Question (QA 292)   - Mail on ALPHA machines

I cannot open Netscape/mail on my AXCAD station. I try and the window keeps dissapearing. This is irritating, please do something about it!

Answer   -   We have already seen "misbehaviours" with Netscape on ALPHA machines. All is reduced to reset your own Netscape configuration which is probably corrupted.

Try this:

 
     cd $HOME 
     mv .netscape .netscape-bad 
 
and then launch Netscape. If "cured", you would only need to copy back both your bookmarks and your addressbook:
 
     cp .netscape-bad/bookmarks.html .netscape 
     cp .netscape-bad/abook.nab .netscape 
     cp .netscape-bad/address-book.html .netscape 
 

Question (QA 298)   - GEM and PEM

If we are encouraged to always use the Generic E-Mail address (GEM), why do the unix phone command and the http://consult.cern.ch/xwho show for quite a few people the Physical E-Mail address (PEM) rather than this GEM.

Answer   -   This depends whether the user uses an account on the CERN mail server or not:
  • If a user has no MAILSERV account at CERN, or doesn't use it for his/her PEM, we recommend to use directly the PEM (usually non-CERN)
  • If a user has a MAILSERV account at CERN and uses it for the PEM (i.e. the PEM has the form xxxxxx@mail.cern.ch), we recommend to use the GEM (usually something like firstname.familynam@cern.ch)

Question (QA 331)   - Mouse support with pine

Any chance to have the mouse support under pine (version 4) - on the CMS Linux cluster?

Answer   -   pine4 has been already introduced in ASIS, but mouse support was introduced already in version 3.95; it is not enabled by default in the CERN configuration. The "pine" parameter to enable via Setup/Config is:
 
 enable-mouse-in-xterm 
 

 


Web Publishing  -  

(Maria Dimou  IT/User Support)

Question (QA 296)   - Network drive \\www\cernweb1%userid

I used to attach the www.cern.ch filesystem as local to my PC's file system by connecting to network drive \\www\cernweb1%userid.

This no more works. Why?

Answer   -   The machine changed name from 'www' to 'web1'. Now the address is \\web1\cernweb1%userid.

You can find full instructions in:

 
 
 http://cern.ch/WebServices/Communications/samba.html 
 

Question (QA 304)   - CGI scripts on the Web

Where should I put my CGI script, so it can be reached through the web?

Answer   -   First you have to register your web site, if you don't have one already. To do this you open the form: http://cern.ch/webregister Thus you obtain the site: http://cern.ch/SITE.

If your files are in afs this corresponds to the directory
/afs/cern.ch/.../SITE/
Create the directory cgi-bin under the directory SITE, i.e.
/afs/cern.ch/.../SITE/cgi-bin/.
You can put your scripts there and they will be web reachable.

Please note: The scripts will only execute one day after the creation of the cgi-bin directory More documentation is available on this subject at http://cern.ch/webservices/AuthoringDoc/Scripting/Scripting.htm

If your CGI needs to read, write or delete a given file in your afs directory, you should also enable the following permissions in your afs ACL:
wwwprot rliw
with the command
fs sa <your_dir_name> wwwprot rliwd

If your files are local to the server web1.cern.ch write to web.support@cern.ch and request your files to be enabled as executables at the level of the web server.


Question (QA 311)   - "Issuer certificate is invalid" with EDH and Netscape

I got problems with EDH web since installing Netscape 4.7. I get (one of) the following errors; after contacting AIS.Support they tell me they cannot help (it is not an EDH but a Netscape problem). Can you help please?

 
     1. # Error: Issuer certificate is invalid (-8156) 
     2. # jarfile: Z\P32\NCOMM47\PROGRAM\java\classes\java40.jar 
     3. # path: Z\P32\NCOMM47\PROGRAM\java\classes\java40.jar 
 
I click on OK and ...
 
     4. Java reported the following error on startup. 
   
     Java.lang.SecurityException : system classes were not signed." 
 

Answer   -   This could mean that your certificates file is corrupted. Close all Netscape windows and delete
 
 c:\netscape\users\<your_login_name>\cert7.db 
 
An empty certificates file will be re-created at the next Netscape start.

About the author(s): Nicole Crémel is working in the User Support group and, with Roger Woolnough, is responsible for the supervision of operations of the Computing Helpdesk. Editing and revising the "Question and Answers" database is a daily task of most members of the group.


For matters related to this article please contact the author.
Cnl.Editor@cern.ch


CERN-CNL-2000-003
Vol. XXXV, issue no 3


Last Updated on Mon Dec 18 13:51:02 GMT+03:30 2000.
Copyright © CERN 2000 -- European Organization for Nuclear Research