On Wed, Nov 11, 2020 at 11:14:33AM +0800, Hu Xuejiao wrote:

[why] When job2sh.rb defined sh_on_fail, it needs to be invoked by the scheduler. Finally, the task is in the waiting state when the task is failed.

The reason is not clear, and it does not explain why it is necessary to wait? Thanks sunyukui

Signed-off-by: Hu XueJiao <1034502035@qq.com> --- sbin/submit | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/sbin/submit b/sbin/submit index bc90e2e0..0bb41f2d 100755 --- a/sbin/submit +++ b/sbin/submit @@ -115,12 +115,14 @@ ARGV.each do |jobfile| sh_run_job = job.sh_run_job sh_extract_stats = job.sh_extract_stats sh_define_files = job.sh_define_files + sh_on_fail = job.sh_on_fail

sh_hash = { 'job2sh' => { 'run_job' => sh_run_job, 'extract_stats' => sh_extract_stats, - 'define_files' => sh_define_files + 'define_files' => sh_define_files, + 'on_fail' => sh_on_fail } }

-- 2.23.0