atmos_main failing with no clear reason

[taubry]

Dear NCAS team,

One of my run failed over the weekend. The suite id is u-bm840. I only made minor modifications since I last ran this suite successfully, and several cycle were run successfully before the suite fail. I thus suspect the problem is related to ARCHER. I checked that I have no disk space/MAUs quota problem.
The last to lines of my job.err file are the following:

[NID 00883] 2019-10-11 23:38:42 Apid 37381831: binary transfer gunzip failed, exit value 1
[FAIL] um-atmos # return-code=1

One of my officemate (Lauren Marshall) has the same problem with suite id u-bm515.
Another officemate (Johnny Staunton Sykes) has a very similar problem (on u-bm865) but the last 2 lines of the job.err show:
[NID 00883] 2019-10-13 19:00:10 Apid 37406754: cannot write placement info file
[FAIL] um-atmos # return-code=1
This seems similar to an ARCHER issue that was supposedly resolved Oct 8th.

For both of them, several cycles were run correctly before failing at atmos_main with these errors.

Thanks for any help!

Thomas

[grenville]

Thomas

This is an ARCHER problem (fixed as far as I know)

NID 00883] 2019-10-13 19:00:10 Apid 37406754: cannot write placement info file

but that's not the problem reported for your job.

Seems like you're running now.

Grenville

[taubry]

Hi Grenville,

Yes we tried to retrigger our jobs this morning and they seem to be running (including those of Lauren and Johnny). Will just check that they don't fail again after a few cycles and close the ticket then.

Thanks,

Thomas