Scaling Up the Workflow on BW

[jacobrh91]

I’ve created the config files necessary to start testing the workflow for 10 samples for both 1 ppn and 2 ppn runs.

Here is the directory I’m currently working in: /scratch/sciteam/jacobrh/purge_exempt/Swift_testing/Justins_Testing/10_sample_set

Justin,

I based the current shell config script based on the more generalized one you created, but I left out the init.sh part (It copies the runfile into the output directory before Swift starts, but this is done at the start of the workflow itself).

Another thing to note is that I had to change this:

PATH=/u/sciteam/wozniak/Public/sfw/compute/swift-t/stc/bin:$PATH
to
PATH=/u/sciteam/wozniak/Public/sfw/login/swift-t/stc/bin:$PATH

As I do not have permission to execute the executable in the compute directory.

Also, when I try to run the test I just created, I get the same error message I received months ago after you had updated the Swift/T installations on BW, it looks like:

Turbine: turbine-cray.sh
 18 01/03/2018 10:10AM
 19 
 20 ENVS
 21 JOB OUTPUT is in /scratch/sciteam/jacobrh/variant-output/X001/output.txt.8000576.bw.out
 22 Running: /u/sciteam/wozniak/Public/sfw/login/tcl-8.6.6/bin/tclsh8.6 swift-t-VariantCalling.NGi.tic -runfile=./1_ppn.runfile
 23 Too many workers allocated to executor types:  {}.^M
 24   Have 0 total workers, allocated^M
 25                   0 already, need at least one to  serve as regular worker^M
 26     while executing^M
 27 "error "Too many workers allocated to executor types:  {$n_workers_by_type}.\n  Have $n_workers total workers, allocated^M
 28                   $workers_run..."^M
 29     (procedure "rank_allocation" line 35)^M
 30     invoked from within^M
 31 "rank_allocation [ adlb::size ] "^M
 32     (procedure "turbine::init" line 17)^M

You were able to fix this last time, so hopefully you have a better idea than I about what is going wrong.