htcondor manager: failure when listening to a telnet commu

[rgavazzi]

See recent comment on the unduly closed issue #107 !

In a nutshell: telnet connection between worker node and master node fails:

telnet: connect to address 192.168.1.3: Connection refused

Is anyone able to run addprocs_htc() on a cluster running htcondor scheduler??
The issue was posted when I was running julia version <=1. 1 but it is still here with v1.4 or v1.5

[aminnj]

Hi, I ran into this issue too. Based on the MPI change mentioned in #107, I made a modification here that allows connections from remote machines

aminnj@f91789b#diff-54c957b90c04bed63e172caa4efa42b072b2e0aef85562ece656d68f8bc8337bL45-R57

In my case, I switched to nc since telnet wasn't available in my worker node environment.
If it works out for you too, I can clean this up and make a PR

[rgavazzi]

Managed to test it finally. It seems to work on my cluster! I still get some erratic connection issues with some particular nodes on the cluster... but this may not be related to ClusterManagers !! I like the additional options, too! Thanks!

[aminnj]

Glad it works for you! :)

[Moelf]

@tanmaykm
probably can close this? and make a new breaking release maybe? for both HTCondor and qsub related overhaul in #153