Gateway timeouts after starting a job

[david-crowder]

I am having issues with running push jobs. Currently, most of the times I start a job it winds up timing out while checking the status. The status seems to be polled more frequently than need be:
https://github.com/chef/knife-push/blob/master/lib/chef/knife/job_start.rb#L98

Is there some way to make knife-push retry when met with a gateway timeout?

[david-crowder]

Looking through the source, it seems like the rest call should be retrying if it hits networking issues:
https://github.com/chef/chef/blob/master/lib/chef/http.rb#L288

[sean-horn]

Zendesk 2641

[markan]

What version of chef/knife are you using? Older versions didn't retry, but the newer versions should, as you pointed out above.

I agree that the polling is too frequent. I'm going to back that off, maybe to 1 sec. When version 2.0 of push server is released we can switch to using the SSE feed and stop the polling entirely.

[david-crowder]

Thanks for looking into this and making the PR @manderson26
The version of knife is 11.18.6. I am excited to see what push server 2.0 will bring.