Troubleshooting

Contents

Introduction

In this page we provide further information and guidelines you need to know about working with VLSI computing resources, along with a collection of some common issues and troubles which can arise working with VLSI lab tools.

xauth: timeout in locking authority file errors

xauth: timeout in locking authority file /users/username/.serverauth.xxx
xauth: timeout in locking authority file /users/username/.Xauthority

Due to UNIX ownership and permissions on .Xauthority file.

Shutdowns and reboots

Machines part of the VLSI cluster are always-on, providing a 24/7/365 service. Moreover, each machine is a multi-user system and different users might have jobs running on the same machine. Thus if you are physically working in the VLSI lab never shutdown yourself a machine at the end of your session work! Just perform a logout instead.
Machine reboots migh be required in case of crashes, hangs or freezes, but always address your issues to the VLSI system administrator before (mazza@NOSPAMto.infn.it, remove NOSPAM in the address).

Global cluster shutdowns might be planned by the system administrator (always before a scheduled electric power interruption to the old building area). In that cases, save your work, close all your running applications and then logout.

Power outage

In case of a power outage you have nothing to do, just wait that the system administrator has restored the cluster.

loss of the electric power to VLSI area

Always check that server machines are reacheable with the ping command,

ping elt59xl.to.infn.it
ping elt95wx.to.infn.it

e controllare i link simbolici in /usr, se il server e' giu' (non riesco a pingarlo) allora i link sono highlighted di rosso e non raggiungibili

ls -l /usr

See also:

reboot_order_after_power_cuts

You can check the state of the licence server or the number of available licences if you run into trouble opening a program.

The lmstat command can be used to track token license activity from the command line

lmstat [-help]

To view the number of available licences for each program,

lmstat -a

Users working remotely from a machine running a fresh installation of Scientific Linux 6.x might need some additional fonts required by Cadence graphical front ends. If you have administration privileges install the missing fonts using the yum package manager,

sudo yum install xorg-x11-font*

or

# yum install xorg-x11-font*

Cadence crashes, hangs or freezes. In general, open a termional and kill all your Cadence processes using the kill command,

ps aux | grep <username>
kill <process id>

For more details see Kill Cadence defunct processes.

Contact the VLSI system administrator (mazza@NOSPAMto.infn.it, remove NOSPAM in the address).

Clean the environment from lock files at startup

When a user opens a cell view for editing (schematic, layout, symbol, etc) Cadence automatically creates a temporary file called a lock file. The lock file prevents another user from opening up the same file for editing at the same time. When one saves and closes the cell view, the lock file is deleted, so that in the future it can be opened for edit again. If one is editing the inverter's layout in cadence, and for some reason cadence crashes or exits for some unexpected reason, the “layout.cdb.cdslck” file may not be deleted, due to the crash.

When this happens, once cadence is restarted, the lock file is still in place. This will make it so you cannot open the inverter's layout. We must remove this lock file manually, as cadence is not intelligent enough on its own to understand what is happening.

This can be accomplished in two ways. From the Cadence IC main window (CIW) you can open the File ⇒ Close Data utility and then select the cell views you want to clean up from lock files and other temporary data:

http://community.cadence.com/cadence_technology_forums/f/38/t/25778


Alternatively you can manually remove the lock files from the UNIX command line. Exit Cadence software and type the following commands in the terminal:

linux% find $HOME/ -name "*.cdslck" -exec rm {} \;
linux% find $HOME/ -name "*.oacache" -exec rm {} \;


You can also cut and paste these instructions into an executable csh script (e.g. ~/scratch/<tech_dir>/cdslk_clean.csh) and call it in the ~/.cshrc to delete all lock files at each login time:

linux% sh ~/scratch/<tech_dir>/cdslk_clean.csh 


Se also

https://secure.engr.oregonstate.edu/wiki/ams/index.php/Cadence/TipsAndTricks
http://www.seas.gwu.edu/~vlsi/ece218/SPRING/reference/tutorial_removing_lock_files.txt
http://www.edaboard.com/thread56960.html


For more information about the Cadence Locking System (CLS), please refer to the official documentation:

Cadence Application Infrastructure User Guide (caiuser.pdf)

Kill Cadence defunct processes

due to a Cadence crash

Each time you logout, kill all process that you have started including “defunct” process. kill -9 -1

or

ps -ef | grep <username>

kill -9 <process id>

zombie process or defunct proces

zombie process

Assicurarsi di non avere processi zombi magari dopo che cadence si pianta all'improvviso!

Visualizzare l'elenco dei processi con ps e killare tutto cio' che riguarda Cadence:

ps aux | grep <username>
kill <process id>

See also

https://intranet.ece.cornell.edu/computing/kb_articles/kba_detail.cfm?article_id=11

Fixing SCP/SFTP

sftp pacher@to01xl.to.infn.it
Connecting to to01xl.to.infn.it...
pacher@to01xl.to.infn.it's password: 
Received message too long 1415934836
scp
...

The problem is related to the local shell environment, not to /etc/ssh/ssh_config SSH server configuration!

http://thomas-cokelaer.info/blog/2011/09/scp-does-not-work-wrong-bashrc/
http://ubuntuforums.org/showthread.php?t=1656869

Solution: don't use echo debugs in a .cshrc or .bashrc file!



Last update: Luca Pacher - May 1, 2013

  • vlsi/workbook/computing/troubleshooting.txt
  • Ultima modifica: 01/02/2015 02:51
  • da pacher