[Imported from gridengine issuezilla ​http://gridengine.sunsource.net/issues/show_bug.cgi?id=3296]

Issue #:  3296
Summary:  Pseudo variable ~ and ~user for qsub -o/-e will crash the execd
Component:  gridengine
Version:  6.2u5 
Platform:  All 
URL:   
OS/Version:  All 
Status:  NEW
Status whiteboard:   
Keywords:   
Resolution:   
Issue type:  DEFECT 
Priority:  P3 
Subcomponent:  clients 
Assigned to:  roland 
Reported by:  reuti

------- Additional comments from reuti@sunsource.net Fri Dec 10 09:14:50 -0800 2010 ------- 

According to the man page of `qsub`, it's valid to use ~ and ~user for the -o
and -e switches. I.e. for the location of the output and error files. Putting this into a job script using either:

#$ -o ~ #$ -o ~reuti

will crash the execd. According to the messages file it's getting a wrong
cryptic username:

12/10/2010 18:06:26| main|pc15370|E|invalid user name
""/gH)ESC^H^\e^]^H250377364337350267@361跨373h233363277yqܷ@361跨373`341^]^H260 233363277300233<U+FF21B>363277

Specifying a file instead of a directory:

#$ -o ~/myoutput

will also not work, as it puts the job in error state, but the execd won't be
killed in this case.

Same is happening when used on the command line while using quotes to avoid
expansion by the shell of these shortcuts.

See ​http://gridengine.org/pipermail/users/2011-August/001448.html et seq.

I just built 8.0.0a on a CentOS 5.6 x86_64 machine

While running the install scripts I ran into a repeatable issue. The install_execd script fails to start the sge_execd daemon - it complains about a missing "./active_jobs" folder inside the directory defined to be the exechost spool directory.

Steps to repeat:

1. Install qmaster normally, I used all the defaults except for choosing classic spooling and not installing the start-on-boot-scripts

2. Run ./install_execd and accept all the default suggestions for execd spoool location. This error will appear:

{{{Grid Engine execution daemon startup

Starting execution daemon. Please wait ...

Shutting down Grid Engine execution daemon

ls: /opt/sge/default/spool/dag/active_jobs: No such file or directory

Hit <RETURN> to continue >>

}}}

Not a fatal error as you can manually go into $CELL/common/ and manually run the sgeexecd start script and things will work fine.

However this may freak out new SGE users/installers