Batch script stdout no longer prints the path to cpl/cesm log file on failure #89
Assignees
Comments
|
I've had the same problem with segfaults not showing up in the log correctly. For what it's worth, my branch is based off of version d1b03ad (not blaming Bill, just demonstrating that the the problem goes back at least to mid-June). It would be nice if the error message was dumped to both stdout and stderr. |
Merged
sholly
pushed a commit
that referenced
this issue
Nov 16, 2015
…nto jedwards4b-moduleloader_fix Fix Moduleloader message, fix logger issues Currently modules are loaded several times in the course of a model build, on yellowstone we are setting an environment variable MP_MPILIB to the internal value of MPILIB, but modules are loaded before this internal variable is set and that was generating an error due to an undefined variable in perl. Change this error to a warning. In the course of this fix I noticed that not all logger messages were making it to stdout. In particular messages from subprograms called with the qx command or with backtics were lost. Changing these calls to system calls fixed this problem. Finally I also converted the check_input_data script to use logger. Test suite: yellowstone driver prealpha suite Test baseline: none Test namelist changes: none Test status: bit for bit Fixes: #304 possibly #89
sholly
pushed a commit
that referenced
this issue
Nov 16, 2015
Fix Moduleloader message, fix logger issues Currently modules are loaded several times in the course of a model build, on yellowstone we are setting an environment variable MP_MPILIB to the internal value of MPILIB, but modules are loaded before this internal variable is set and that was generating an error due to an undefined variable in perl. Change this error to a warning. In the course of this fix I noticed that not all logger messages were making it to stdout. In particular messages from subprograms called with the qx command or with backtics were lost. Changing these calls to system calls fixed this problem. Finally I also converted the check_input_data script to use logger. Test suite: yellowstone driver prealpha suite Test baseline: none Test namelist changes: none Test status: bit for bit Fixes: #304 possibly #89
|
fixed in #307 |
jedwards4b
pushed a commit
to jedwards4b/cime
that referenced
this issue
May 31, 2019
…6c79 Jedwards/mergetob896c79
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
As of beta06 (with updated CIME and CAM tags), if I have a CESM run crash (e.g. by segfaulting), I'm no longer seeing a message printing whether the run failed, with the log file location. The last stdout message is simply:
(To make this message easier to find, it should be in stdout rather than stderr I think, but for the record there's nothing printed to stderr, either.)
I see that the batch script still has Perl logic to do this, including this line:
However, there must be something wrong with the logic used to reach this point, since the log file exists, but nothing is printed.
The text was updated successfully, but these errors were encountered: