[HTCondor-users] condor_ssh_to_job fails in 9.0.17

[Vikrant Aggarwal <ervikrant06@xxxxxxxxx>]

Hello Experts,

We are flocking jobs from on-prem to cloud, since we are working on upgrading from 8.8.5 to 9.0.17. Currently using host based security in 9.0.17 to make this migration smooth.Â

Test case 1: Submit job from 9.0.17 on-prem submit node on cloud setup running 8.8.5. (I have to add CLAIMTOBEÂsecurity method to make comm with cloud) Job started running on cloud worker node but condor_ssh_to_job from on-prem to cloud is failing. BTW it's working with an on-prem 8.8.5 submit node.Â

Debug option of condor_ssh_to_job shows submitÂnode is trying to do ssh with nobody user instead of the actual user ID. It's confirmed that the cloud job is running with the right user ID.Â

01/09/23 07:07:27 (fd:5) (pid:635869) (D_ALWAYS:2) Executing ssh command: ssh -oUser=nobody -oIdentityFile=/tmp/testuser1.condor_ssh_to_job_fbdafb26/ssh_key -oStrictHostKeyChecking=yes -oUserKnownHostsFile=/tmp/testuser1.condor_ssh_to_job_fbdafb26/known_hosts -oGlobalKnownHostsFile=/tmp/testuser1.condor_ssh_to_job_fbdafb26/known_hosts -oProxyCommand="condor_ssh_to_job"' '"-debug"' '"-proxy"' '"/tmp/testuser1.condor_ssh_to_job_fbdafb26/fdpass" test.cloud.example.com

Everything with 8.8.5 works as expected, gone through release notes but couldn't see anything which can change this behavior in 9.0.17 version.Â

Thanks & Regards,

Vikrant Aggarwal