Skip to content
This repository has been archived by the owner on Mar 16, 2022. It is now read-only.

newly created directories not always available fro remote job #49

Closed
flowers9 opened this issue May 31, 2017 · 1 comment
Closed

newly created directories not always available fro remote job #49

flowers9 opened this issue May 31, 2017 · 1 comment

Comments

@flowers9
Copy link

This is similar to an earlier issue I commented on, and still appears to be an issue. The wrapper script created by simple_pwatcher_bridge.py tries to cd {wdir}, and it fails because that directory hasn't yet propogated from the program's head node to the remote node the job is run on. This one has a very simple fix:

--- simple_pwatcher_bridge.py 2017-05-31 10:57:47.317485000 -0500
+++ simple_pwatcher_bridge.py 2017-05-31 10:58:59.469657000 -0500
@@ -350,6 +350,7 @@
rel_actual_script_fn = os.path.relpath(actual_script_fn, wdir)
wrapper = """#!/bin/sh
set -vex
+mkdir -p {wdir}
cd {wdir}
bash {rel_actual_script_fn}
touch {sentinel_done_fn}

mkdir -p will return success if the directory already exists, and the attempt to create the directory forces it to propogate to the node the job is running on.

@pb-cdunn
Copy link

The wrappers (which run from the installation, not from the remote directory) should now cause the remote job to wait for all known input files to become available.

pb-cdunn pushed a commit that referenced this issue Aug 28, 2018
…develop

* commit '85c6d6e8b70145daa600bb29b3b4e3a32e1d3735':
  Try again to fix MetaJobs (fs_based pwatcher)
Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants