[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [HTCondor-users] Simplest way to change ulimit for all new jobs

Hello Todd,

I will therefore use your suggestion which brings more protection. But it will require to drain the exec nodes first.



Le 04/10/2018 19:44, Todd Tannenbaum a écrit :
On 10/4/2018 7:42 AM, Christophe DIARRA wrote:
Hello Sean,

I have used an advice from one colleague (Andrea Sartirana
<sartiran@xxxxxxxxxxxx>). The idea is to use the USER_JOB_WRAPPER variable.

Hi Christophe,

I strongly suggest the init / systemd based mitigation I just posted to this group (copied again below for convenience) instead of a USER_JOB_WRAPPER approach.

The reason is the USER_JOB_WRAPPER is literally only applied to the job itself, and will not be applied to job pre or post scripts (that can come from the user), ssh_to_job sessions, DAGMan pre/post scripts, or scheduler universe jobs.

** For HTCondor v8.6.x+ installed from RPMs on RHEL6, SL6, Centos6 (i.e. distros running init):

As root run the following commands:

    sed -i 's/ULIMIT_FLAGS=.*/ULIMIT_FLAGS="-Hs 16000000"/' /etc/sysconfig/condor
    service condor restart

** For HTCondor v8.6.x+ installed from RPMs or DEBs on RHEL7, Centos7, SL7, Debian, or Ubuntu (i.e. distros running systemd):

As root run the following commands:

   mkdir /etc/systemd/system/condor.service.d
   echo -e '[Service]\nLimitSTACK=16G\n' > /etc/systemd/system/condor.service.d/CVE-2018-14634.conf
   systemctl restart condor

Christophe DIARRA
Institut de Physique Nucleaire
15 Rue Georges Clemenceau
S2I/D2I - Bat 100A - Piece A108
F91406 ORSAY Cedex
Tel:    +33 (0)1 69 15 65 60 / +33 (0)6 31 26 23 69
Fax:    +33 (0)1 69 15 64 70 / E-mail: diarra@xxxxxxxxxxxxx