Did you manage to figure out whether the ssh connection is dropped or
the `oc port-forward` connection?
I think you can use wireshark to watch the `oc` remote connection and
see when it gets dropped and by which side of it.
There are a lot of things that can be tried. For example setting
connection keepalives.
Also (since you are using `oc port-forward`), maybe you can base your
pods on the `origin-ansible` image and use `oc exec mypod -- ansible
...` instead of calling local ansible through port forward?
It is good to first understand though where the connection drop happens.
P.S. I've seen such disconnects on virtual machines that have nothing to
do with OpenShift. Searching for understanding at the time didn't yield
any results except for finding some other reports. My issue got fixed by
itself (I don't remember how) and never hit it again.
Fabio Martinelli wrote on 04/19/18 11:46:
Dear Colleagues
In few time I've to migrate several corporate applications from a
RedHat6 LXC cluster to a RedHat7 OpenShift Origin 3.7.2 cluster
here the application Developers are use to write an Ansible playbook for
each app so they've explicitly requested me to prepare a base CentOS7
container running as non-root and featuring an unprivileged SSHd daemon
in order to run their well tested Ansible playbooks, furthermore to
place the container /home on a dedicated GlusterFS volume to make it
persistent along the time ; last ring of this chain is the oc
port-forward command that's in charge of connecting the Developers
workstation with the unprivileged SSHd daemon just for the Ansible
playbook execution time.
this is actually working pretty well but the fact that the oc
port-forward command at certain point cuts the connection and the
Ansible run gets obviously affected making the Developer experience
disappointing ; on the other end the SSHd process didn't stop.
kindly which settings may I change both on the Origin Masters yaml files
and on the Origin Nodes yaml files in order to prevent this issue ?
I'm aware that the application Developers should rewrite their works in
terms of Dockerfiles but for the time being they've really no time to do
that.
Many thanks,
Fabio Martinelli
_______________________________________________
users mailing list
[email protected]
http://lists.openshift.redhat.com/openshiftmm/listinfo/users
_______________________________________________
users mailing list
[email protected]
http://lists.openshift.redhat.com/openshiftmm/listinfo/users
