condor_submit_workers(1)

📦
version 9.9 · 3 versions
, in debian bookworm, ubuntu jammy, ubuntu noble

💅 condor_submit_workers — the manual page, bestie edition

okay so you opened condor_submit_workers(1) from the Cooperative Computing Tools and you want the tea. you're in the right place. sit down, grab a strawberry, let's go 🍓

✨ NAME ✨

condor_submit_workers — submit work_queue_worker to the Condor grid. that's it, that's the whole job. no cap. 💫

🌸 SYNOPSIS 🌸

condor_submit_workers [options] <servername> <port> <num-workers>

or

condor_submit_workers [options] --manager-name <name> <num-workers>

two ways to do the thing. pick your fighter 🥊

💖 DESCRIPTION 💖

so condor_submit_workers schedules the execution of work_queue_worker(1) on a grid that Condor runs, and it does it through Condor's own job submission interface, condor_submit. iconic behaviour honestly.

the number of work_queue_worker that get scheduled and run? that's literally the num-workers argument. you say 10, you get 10. periodt. 👑

the servername and port arguments tell it the hostname and the port number of the manager that the work_queue_worker should connect to. but — plot twist — you don't have to know that. you can get that info from the catalog server instead by giving the name of the work queue with the --manager-name parameter. flexible queen 🌈

👑 OPTIONS 👑

one bullet per flag, zero skips, we do not leave the group chat on read.

  • -M, --manager-name=<name> — the name of the preferred manager for the worker. your fave. 💕
  • -N, --name=<name> — literally the same as -M, it's just here for backwards compatibility. vintage behaviour, we respect it.
  • -C, --catalog=<catalog> — set the catalog server to <catalog>. the <catalog> format is HOSTNAME:PORT. yes, colon, no spaces, don't make it weird.
  • -t, --timeout=<time> — abort after this much idle time. default is 900s. if it's just sitting there doing nothing, it's outta here 🚪
  • -d, --debug=<subsystem> — turn on debugging on the worker for this subsystem. not sure where to start? try -d all. go feral, bestie.
  • -w, --tcp-window-size=<size> — set the TCP window size.
  • -i, --min-backoff=<time> — the initial value for the backoff interval, for when the worker fails to connect to a manager. default is 1s. it starts small, it's humble.
  • -b, --max-backoff=<time> — the maximum value for that backoff interval, same situation, worker can't reach a manager. default is 60s. it stops growing here, boundaries ✋
  • -z, --disk-threshold=<size> — set the available disk space threshold in MB. when it gets exceeded, the worker will clean up and reconnect. default is 100MB. tidy little thing 🧹
  • -A, --arch=<arch> — set the architecture string that the worker reports to the manager, instead of whatever value comes from uname.
  • -O, --os=<os> — set the operating system string that the worker reports to the manager, instead of whatever value comes from uname.
  • -s, --workdir=<path> — set the location where the working directory of the worker gets created.
  • -P, --password=<file> — password file used to authenticate workers to the manager. keep it secret, keep it safe 🔐
  • -E, --worker-options=<str> — extra options passed straight through to work_queue_worker.
  • --cores <cores> — set how many cores each worker should use (0 means auto). default is 1.
  • --memory <size> — manually set the amount of memory (in MB) that this worker reports.
  • --disk <size> — manually set the amount of disk (in MB) that this worker reports.
  • -r, --requirements=<reqs> — the Condor requirements expression.
  • --class-ad <ad> — an extra Condor class ad. you can specify this one multiple times, stack 'em up 🥞
  • --autosize — Condor will automatically size the worker to the slot. self-sizing, main character energy.
  • --docker-universe <image> — run the worker inside <image> using Condor's docker universe 🐳
  • -h, --help — show the help message.

💀 EXIT STATUS 💀

  • on success, it returns zero. slay ✅
  • on failure, it returns non-zero. not slay, but at least it's honest ❌

🍓 EXAMPLES 🍓

Example the first: submit 10 worker instances to run on Condor, connecting to one specific manager:

condor_submit_workers manager.somewhere.edu 9123 10

Example the second: submit 10 work_queue_worker instances to run on Condor in auto mode, with their preferred project name set to Project_A and the abort timeout set to 3600 seconds:

condor_submit_workers -a -t 3600 -M Project_A 10

The Cooperative Computing Tools are Copyright (C) 2005-2019 The University of Notre Dame. This software is distributed under the GNU General Public License. See the file COPYING for details.

🌈 SEE ALSO 🌈

okay that's the whole page, you ate that 🫶 this was the condor_submit_workers(1) page from CCTools 8.0.0 DEVELOPMENT — version 8.0.0, development era, the manual name at the top, the manual name at the bottom, the whole cute little sandwich. mwah. 💖✨🎀🦋

📦 where this page comes from