![]() If you cannot run your application interactively because your job requests a large number of nodes or it takes a long time to reach a problematic area, the above interactive approach is not practical. The above example is to use SIGABRT in killing the application. ATP analysis proceeding.ĪTP Stack walkback for Rank 0 Stack walkback for Rank 0 done $ scancel -s ABRT 3169879.0 # Kill the applicationĪpplication 3169879 is crashing. $ sacct -j 3169879 # find job step id for the application - it's 3169879.0 JobID JobName Partition Account AllocCPUS State ExitCodeģ169879.ext+ extern mpccc 544 RUNNING 0:0 To view the collected backtrace result, you need to load the stat module on Cori or cray-stat on Perlmutter, and run stat-view: Srun: Force Terminated job step 3044170.0ĪTP creates a merged stack backtrace files in DOT fomat in atpMergedBT.dot (with function-level aggregation) and atpMergedBT_line.dot (with line-level aggregation). View application merged backtrace tree with: STATview atpMergedBT.dot Process died with signal 4: 'Illegal instruction' ATP analysis proceeding.ĪTP Stack walkback for Rank 3 Stack walkback for Rank 3 done #!/bin/bash #SBATCH -N 1 #SBATCH -t 5:00 #SBATCH -q debug export ATP_ENABLED = 1Īpplication 3044170 is crashing.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |