Major Hello everybody,
Since the latest LTS update we are experiencing a lot of http queries that being cut off by jenkins (everything worked fine until the update).
Resulting to 502 errors from our nginx reverse proxy.
To reproduce the error I loop on the jenkins host (until it fails) with the given command:
while curl -o /dev/null -v http://localhost:8080/login -s ; do date; echo '######################'; sleep 1 ; date; done ; date
When it errors (some seconds to a couple of minutes) we get:
Wed Sep 19 11:14:46 CEST 2018
* Hostname was NOT found in DNS cache
* Trying 127.0.0.1...
* Connected to localhost (127.0.0.1) port 8080 (#0)
> GET /login HTTP/1.1
> User-Agent: curl/7.38.0
> Host: localhost:8080
> Accept: /
>
* Empty reply from server
* Connection #0 to host localhost left intact
Wed Sep 19 11:14:52 CEST 2018
There is always 5 to 6 seconds between the query and the cut-off.
There is no log relevant (some scm polling only or nothing at all when the error occurs).
This could append on any query, here login, but also when scrolling the console, or any other page / API call.
