pavilion2
pavilion2 copied to clipboard
hpc
Pavilion is a Python 3 (3.5+) based framework for running and analyzing tests targeting HPC systems.
On the same level as 'action'. This avoids printing results with an absurd number of digits to the terminal. And facilitates displaying results as the user desires. Code review checklist:...
Build on a node and the build fails but the status file looks like: ``` bash 1712858170.457788 STATUS_CREATED Created status file. 1712858170.458920 CREATED Test directory and status file created. 1712858170.637588...
Our succinct errors are nice, except when they're not informative enough to get us to the problem. Add --debug. It should force the error handling in main.py and errors.pformat() to...
Series should periodically check node status in for tests with prescribed nodes, and cancel tests if node are no longer up.
The name of `sched.test_cmd` has bothered me for a long time. It's time to change it. Rename to `sched.mpi_cmd` and make `sched.test_cmd` an alias to that. Update documentation and examples...
Add the ability to send result logs to multiple locations.
``` pav results -kname,mpi_ranks,slowest_rank,runtime_seconds,network_timeouts,minutes "$@" Test Results for: s1154 -------+---------------------+----------+--------+---------------------+-----------+--------------+-----------------+------------------+--------- Id | Name | Started | Result | Name | Mpi ranks | Slowest rank | Runtime seconds | Network...
``` source_url: source_download: missing ``` does not end up downloading the source Also, when `source_download` is set to `latest`, I get ``` ... Could not create builder for test stream.base...
Within a test set, we need to be able to interleave tests. This will require a new 'interleave' option in series (at both the top and per test_set level). The...
This requires some discussion, but the jist is: We need to point at builds that we don't actually want to copy. I'm thinking this should have two components: 1. An...