Re: System hang problem.

We solved precisely this problem on our interactive login systems.  We 
wanted to limit anybody that logged in to 3GB of useable memory (process 
size, not RSS).

To do this, I added the following two scripts to /etc/profile.d:

[skaven@adcdog03 /etc/profile.d]$ cat limits.csh
# Set a 3GB process limit size
limit vmemoryuse 3096000
[skaven@adcdog03 /etc/profile.d]$ cat limits.sh
# Set a 3GB process limit size
ulimit -v 3096000

I then had to modify /etc/csh.cshrc and /etc/csh.login.  For some odd 
reason the RHEL folks have the /etc/profile.d stuff executing for login 
shells for bash, but not tcsh.  So I added this little tidbit to our 
cfengine config:

# Fix RHEL3/RHEL4 /etc/profile.d/*.csh issues -- profile.d should
# only be read for login shells for both bash and tcsh.
profile.(RHEL3|RHEL4)::

# Remove profile.d stuff from cshrc
{ /etc/csh.cshrc
     LocateLineMatching "^if.*/etc/profile.d.*"
     DeleteToLineMatching "^endif$"
     DeleteNLines "1"
     CatchAbort
}

# Add profile.d stuff to login
{ /etc/csh.login
     BeginGroupIfNoLineMatching "^if.*/etc/profile.d.*"
         Append ""
         Append "if ( -d /etc/profile.d ) then"
         Append "${tab}set nonomatch"
         Append "${tab}${tab}foreach i ( /etc/profile.d/*.csh )"
         Append "${tab}${tab}if ( -r $i ) then"
         Append "${tab}${tab}${tab}source $i"
         Append "${tab}${tab}endif"
         Append "${tab}${tab}end"
         Append "${tab}unset i nonomatch"
         Append "endif"
     EndGroup
}


The end result is that anybody logging into one of these systems has a 
hard limit of 3GB applied to their shell.  If you want this limit to 
apply to non-login shells, you can try modifying /etc/csh.cshrc and 
/etc/bashrc to put the limits in there, though I haven't tried to see if 
that works or not.

Paul Krizak                         5900 E. Ben White Blvd. MS 625
Advanced Micro Devices              Austin, TX  78741
Linux/Unix Systems Engineering      Phone: (512) 602-8775
Silicon Design Division


Manish Neema wrote:
> It is not uncommon for us to see a process size ~40GB on 16 GB machines.
> I know, the simple answer would be to buy machines with more RAM.
> 
> Anyway, I'll ask my last question again. Under the given circumstances,
> is there any way for me to limit the jobs based on SIZE (and NOT RSS;
> their RAM footprint)?
> 
> -Manish
> 
> 
> 
> --
> Taroon-list mailing list
> [email protected]
> https://www.redhat.com/mailman/listinfo/taroon-list
> 
> 


--
Taroon-list mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/taroon-list