Hi,

I think it is worth to add a new rule based on a process name.
I'd like to know everyone's opinion. If no objection, I will work
for implementing this rule.
Any comment is welcome.


Requirement:
============
The existing rule based on UID/GID is worth in many cases, that an
administrator prepares an exclusive user for each program (apache,
postgresql, etc.).
In some cases, this rule is not enough. For example, some backup
tool runs as root user because it needs to access any disk for the
backup, and it uses a lot of memory. So we'd like to restrict its
memory usage automatically, but the rule based on UID/GID can not
be used.
So I think it is worth to add a new rule based on a process name.


A NEW RULE (/etc/cgrules.conf):
===============================
CURRENT) <user>                 <controllers>   <destination>
NEW)     <user>:<process name>  <controllers>   <destination>

I think the current rule should be used for the compatibility of
course. In addition a new rule is like the following:
  <user>:<process name>  <controllers>   <destination>

If matching both process's user and process name with a rule when
an EXEC event happens, the process is moved to cgroup <destination>
of subsystem <controllers>.


Thanks
Ken'ichi Ohmichi

------------------------------------------------------------------------------
Crystal Reports - New Free Runtime and 30 Day Trial
Check out the new simplified licensing option that enables 
unlimited royalty-free distribution of the report engine 
for externally facing server and web deployment. 
http://p.sf.net/sfu/businessobjects
_______________________________________________
Libcg-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libcg-devel

Reply via email to