Uploaded image for project: 'WiredTiger'
  1. WiredTiger
  2. WT-8367

PPC format failures pid <number> is not a child of this shell

    • 5
    • Storage - Ra 2022-01-24

      While looking at a random format stress failure, in the output log every process started is resulting in a line:

      [2021/11/09 06:54:12.114] format.sh: starting job in /data/mci/50eeff57c8a356cfb82184078ee6a674/wiredtiger/test/format/RUNDIR.167 (Tue Nov  9 06:54:12 UTC 2021)
      [2021/11/09 06:54:12.114] format.sh:  catchsegv ./t -c /data/mci/50eeff57c8a356cfb82184078ee6a674/wiredtiger/test/format/CONFIG.stress -h /data/mci/50eeff57c8a356cfb82184078ee6a674/wiredtiger/test/format/RUNDIR.167   quiet=1
      [2021/11/09 06:54:59.243] ./format.sh: line 412: wait: pid #### is not a child of this shell

      This happens on every starting job in... loop iteration. Is there a bug in format.sh where it is not properly storing the child PID? Is there a way to catch or fix those messages. They clutter the output and distract from the real issue.

      Secondly, at the end there are nearly a hundred lines like this:

      [2021/11/09 07:12:34.187] ./format.sh: line 404: kill: (-16669) - No such process
      [2021/11/09 07:12:34.187] ./format.sh: line 405: wait: pid 16669 is not a child of this shell
      [2021/11/09 07:12:34.410] ./format.sh: line 404: kill: (-4353) - No such process
      [2021/11/09 07:12:34.410] ./format.sh: line 405: wait: pid 4353 is not a child of this shell

      Again, it would be nice to catch these and not clutter the real failure.

      These examples were taken from https://evergreen.mongodb.com/task_log_raw/wiredtiger_ubuntu1804_ppc_format_stress_ppc_zseries_test_28109e1c6a2444d1ebcf8ef87ddc6944cf88eb61_21_11_09_04_48_07/0?type=T&text=true

            keith.bostic@mongodb.com Keith Bostic (Inactive)
            sue.loverso@mongodb.com Susan LoVerso
            0 Vote for this issue
            4 Start watching this issue