arch task returns 0 when htar fails

[AndrewEichmann-NOAA]

What is wrong?

The arch task is returning 0 and marked by rocoto has having completed successfully when htar fails and returns a non-zero status. Example from log below:

+ exglobal_archive.sh[265]: htar -P -cvf /NCEPDEV/marineda/1year/Andrew.Eichmann/HERA/scratch/arch-test/2021032418/gdas_restartb.tar gdas.20210324/18//model_data/atmos/restart
[connecting to hpsscore1.fairmont.rdhpcs.noaa.gov/1217]
ERROR: htar_MakeAllSubdirs: Permission error creating intermediate HPSS subdirectory: /NCEPDEV/marineda/1year/Andrew.Eichmann
HTAR: HTAR FAILED
###WARNING  htar returned non-zero exit status.
            72 = /apps/hpss/bin/htar -P -cvf /NCEPDEV/marineda/1year/Andrew.Eichmann/HERA/scratch/arch-test/2021032418/gdas_restartb.tar gdas.20210324/18//model_data/atmos/restart
+ exglobal_archive.sh[266]: status=72
+ exglobal_archive.sh[267]: case ${targrp} in
+ exglobal_archive.sh[274]: '[' 72 -ne 0 ']'
+ exglobal_archive.sh[274]: '[' 2021032418 -ge 2021032512 ']'
+ exglobal_archive.sh[278]: set_strict
+ preamble.sh[41]: [[ YES == \Y\E\S ]]
+ preamble.sh[43]: set -eu
+ exglobal_archive.sh[281]: shopt -u extglob
+ exglobal_archive.sh[287]: exit 0
+ exglobal_archive.sh[1]: postamble exglobal_archive.sh 1698268242 0
+ preamble.sh[68]: set +x
End exglobal_archive.sh at 21:11:20 with error code 0 (time elapsed: 00:00:38)
+ JGLOBAL_ARCHIVE[32]: status=0
+ JGLOBAL_ARCHIVE[33]: [[ 0 -ne 0 ]]
+ JGLOBAL_ARCHIVE[42]: [[ -e OUTPUT.111163 ]]
+ JGLOBAL_ARCHIVE[50]: cd /scratch1/NCEPDEV/stmp2/Andrew.Eichmann/RUNDIRS/arch-test
+ JGLOBAL_ARCHIVE[51]: [[ NO = \N\O ]]
+ JGLOBAL_ARCHIVE[51]: rm -rf /scratch1/NCEPDEV/stmp2/Andrew.Eichmann/RUNDIRS/arch-test/arch.110645
+ JGLOBAL_ARCHIVE[53]: exit 0
+ JGLOBAL_ARCHIVE[1]: postamble JGLOBAL_ARCHIVE 1698268240 0
+ preamble.sh[68]: set +x
End JGLOBAL_ARCHIVE at 21:11:20 with error code 0 (time elapsed: 00:00:40)
+ arch.sh[17]: status=0
+ arch.sh[19]: exit 0
+ arch.sh[1]: postamble arch.sh 1698268238 0
+ preamble.sh[68]: set +x
End arch.sh at 21:11:21 with error code 0 (time elapsed: 00:00:43)

---

What should have happened?

I would expect the arch task to fail when any htar command within it fails.

What machines are impacted?

Hera

Steps to reproduce

1. Run a cycling experiment with HPSS_PROJECT in config.base set to something the user has no access to
2. Let the arch task run - the first one appears to suffice

Additional information

Is this behavior in fact expected?

Do you have a proposed solution?

No response

[WalterKolczynski-NOAA]

@AndrewEichmann-NOAA can you test again? I think this may be been fixed as a side effect of #1967, which was merged this afternoon.

[AndrewEichmann-NOAA]

@WalterKolczynski-NOAA Yup it works now. I mean it doesn't work. In the right way.