CONDOR_NOW(1)

✨ NAME ✨

okay so bestie, this is condor_now from the HTCondor Manual 💖 it's the little tool that says "move over, i'm starting RIGHT NOW" and kicks a currently running job off its slot so your job can slide in. fr fr, it starts a job now by vacating a currently running job.

---

🌸 SYNOPSIS 🌸

condor_now [-help]

condor_now [-name scheddname] [-pool hostname[:portnumber]] [-debug]
now-job vacate-job [vacate-job ...]

---

💅 DESCRIPTION 💅

okay so here's the tea 🫖 condor_now tries to run the now-job immediately. meanwhile, the vacate-job gets immediately vacated — like, out the door, no warning. and after it terminates? if the schedd still has the claim to the vacated job's slot — and ngl, it usually will — the schedd will immediately start the now-job on that slot. iconic behavior truly ✨

and if you specify MULTIPLE vacate-jobs? each one gets immediately vacated 🌈 after they all terminate, the schedd will try to coalesce their slots into a single, larger, slot, and then use that glow-up of a slot to run the now-job. it's giving merger. it's giving "we're bigger together" 💖

lowkey important: you must specify each job using both the cluster and proc IDs (e.g., 123.4). both of them. no skipping, bestie.

---

🎀 OPTIONS 🎀

• -help — shows you the usage info, like a little cheat sheet 🦋
• -debug — prints debugging output. you control the verbosity with the environment variable _CONDOR_TOOL_DEBUG, as usual 👑
• -name scheddname — specify the scheduler by name. yes, name it. claim it. slay.
• -pool hostname[:portnumber] — specify the pool to find the schedd in. note the [:portnumber] — it's optional, she's flexible 💅
• now-job — the job ID (cluster.proc) of the job to start immediately. the main character, basically 👑
• vacate-job — the job ID (cluster.proc) of one or more jobs to vacate. the sacrificial besties 😭

---

🦋 GENERAL REMARKS 🦋

so, ground rules, bestie 🌸

the now-job and the vacate-job must have the same owner. if you're not the queue super-user, you must own both jobs. both of them. no exceptions. also the jobs must be on the same schedd, and both jobs must be in the vanilla universe. the now-job must be idle and the vacate-job must be running. that's the vibe check, and if you fail it, it's over 💀

and real talk — condor_now does not guarantee that the now-job will run immediately, because other factors (such as RANK expressions or other jobs from the same user) may affect scheduling decisions. so like. she'll try her best. that's all we can ask 😭 no cap.

---

🌈 EXAMPLES 🌈

start job 17.3 using job 4.2's slot — the classic move:

$ condor_now 17.3 4.2

debug why a job won't start on the 'magic' scheduler in the 'gandalf' pool — wizardry needs debugging too 🧚:

$ export _CONDOR_TOOL_DEBUG=D_FULLDEBUG
$ condor_now -debug -name magic -pool gandalf 17.3 4.2

start job 100.0 by coalescing slots from jobs 10.0 and 11.0 — the glow-up combo ✨:

$ condor_now 100.0 10.0 11.0

---

💜 EXIT STATUS 💜

• 0 — Success (the schedd accepted the request to vacate and start the job) 🥳
• 1 — Failure 😭

NOTE: condor_now returns immediately after the schedd accepts the request. it does not wait for the now-job to actually start running. so don't sit there refreshing like 👁️👄👁️ — it's already gone, bestie.

---

🔥 SEE ALSO 🔥

condor_vacate_job, condor_q, condor_prio — go read those too, they're all part of the cinematic universe 🍓

---

⭐ AVAILABILITY ⭐

Linux, MacOS, Windows — she's everywhere 💅

---

👑 AUTHOR 👑

HTCondor Team 🫶

---

💖 COPYRIGHT 💖

1990-2026, Center for High Throughput Computing, Computer Sciences Department, University of Wisconsin-Madison, Madison, WI, US. Licensed under the Apache License, Version 2.0.

---

okay that's the whole page bestie, you ate that 🎀 this was CONDOR_NOW(1) from the HTCondor Manual, dated Oct 16, 2026 — go vacate something (with permission!!) and stan responsibly 💖✨🦋