diff --git a/NEWS b/NEWS index f788d7802cd5ef7d9e525a1dadca0a6aadc02f41..60c50471a9ab11203e4fc37d62ad459162ff8a13 100644 --- a/NEWS +++ b/NEWS @@ -117,6 +117,8 @@ documents those changes that are of interest to users and administrators. -- Fix priority/multifactor priorities on a slurmctld restart if not using accounting_storage/[mysql|slurmdbd]. -- Fix NULL dereference reported by CLANG. + -- Update proctrack documentation to strongly encourage use of + proctrack/cgroup. * Changes in Slurm 17.02.2 ========================== diff --git a/doc/html/configurator.easy.html.in b/doc/html/configurator.easy.html.in index 2cb86ab4edef222958ed5c95e21c84b9e3491d55..19e6b84458114cc1059df98a013ffa6f9480796a 100644 --- a/doc/html/configurator.easy.html.in +++ b/doc/html/configurator.easy.html.in @@ -380,11 +380,11 @@ Define the algorithm used to identify which processes are associated with a given job. This is used signal, kill, and account for the processes associated with a job step.<BR> Select one value for <B>ProctrackType</B>:<BR> -<input type="radio" name="proctrack_type" value="cgroup"> <B>Cgroup</B>: Use +<input type="radio" name="proctrack_type" value="cgroup" checked> <B>Cgroup</B>: Use Linux <i>cgroup</i> to create a job container and track processes. Build a <i>cgroup.conf</i> file as well<BR> <input type="radio" name="proctrack_type" value="cray"> <B>Cray</B>: Cray proprietary process tracking<BR> -<input type="radio" name="proctrack_type" value="pgid" checked> <B>Pgid</B>: Use Unix +<input type="radio" name="proctrack_type" value="pgid"> <B>Pgid</B>: Use Unix Process Group ID, processes changing their process group ID can escape from Slurm control<BR> <input type="radio" name="proctrack_type" value="linuxproc"> <B>LinuxProc</B>: Use diff --git a/doc/html/configurator.html.in b/doc/html/configurator.html.in index 147b0079138dd149b16277967e1e72b399b4471d..f6de8f690e05bf98bbaa9159a502a3b722c4d089 100644 --- a/doc/html/configurator.html.in +++ b/doc/html/configurator.html.in @@ -533,11 +533,11 @@ Define the algorithm used to identify which processes are associated with a given job. This is used signal, kill, and account for the processes associated with a job step.<BR> Select one value for <B>ProctrackType</B>:<BR> -<input type="radio" name="proctrack_type" value="cgroup"> <B>Cgroup</B>: Use +<input type="radio" name="proctrack_type" value="cgroup" checked> <B>Cgroup</B>: Use Linux <i>cgroups</i> to create a job container and track processes. Build a <i>cgroup.conf</i> file as well<BR> <input type="radio" name="proctrack_type" value="cray"> <B>Cray</B>: Cray proprietary process tracking<BR> -<input type="radio" name="proctrack_type" value="pgid" checked> <B>Pgid</B>: Use Unix +<input type="radio" name="proctrack_type" value="pgid"> <B>Pgid</B>: Use Unix Process Group ID, processes changing their process group ID can escape from Slurm control<BR> <input type="radio" name="proctrack_type" value="linuxproc"> <B>LinuxProc</B>: Use diff --git a/doc/html/high_throughput.shtml b/doc/html/high_throughput.shtml index 404f4892cc4784e0cd96c76c1cb29c3e0c629d14..92f5747d52644150b14d3f14b939da67d6e0f82f 100644 --- a/doc/html/high_throughput.shtml +++ b/doc/html/high_throughput.shtml @@ -118,9 +118,6 @@ The default value is 300 seconds. The value should be reduced to a few seconds if possible. Use of accounting records for older jobs can increase the job throughput rate compared with retaining old jobs in the memory of the slurmctld daemon.</li> -<li><b>ProctrackType</b>: -Avoid using <i>proctrack/cgroup</i> it is considerably slower than -other alternatives. <i>proctrack/pgid</i> (the default) is advised.</li> <li><b>PriorityType</b>: The <i>priority/builtin</i> is considerably faster than other options, but schedules jobs only on a First In First Out (FIFO) basis.</li> diff --git a/doc/man/man5/slurm.conf.5 b/doc/man/man5/slurm.conf.5 index c9d4eee507f7c5be685bd4946546f4d7e528cb7e..059a8d590fcdbd867267528d9f1e80d8b2826b50 100644 --- a/doc/man/man5/slurm.conf.5 +++ b/doc/man/man5/slurm.conf.5 @@ -2005,8 +2005,7 @@ NOTE: "proctrack/linuxproc" and "proctrack/pgid" can fail to identify all processes associated with a job since processes can become a child of the init process (when the parent process terminates) or change their process group. -To reliably track all processes, one of the other mechanisms -utilizing kernel modifications is preferable. +To reliably track all processes, "proctrack/cgroup" is highly recommended. NOTE: The \fBJobContainerType\fR applies to a job allocation, while \fBProctrackType\fR applies to job steps. Acceptable values at present include: @@ -2015,10 +2014,6 @@ Acceptable values at present include: \fBproctrack/cgroup\fR which uses linux cgroups to constrain and track processes. NOTE: see "man cgroup.conf" for configuration details -NOTE: This plugin writes to disk often and can impact performance. -If you are running lots of short running jobs -(less than a couple of seconds) this plugin slows down performance dramatically. -It should probably be avoided in an HTC environment. .TP \fBproctrack/cray\fR which uses Cray proprietary process tracking