Re: running health checks from behind a web proxy

Robert Marshall <robert.marshall@...>

Thanks - hope it works for you, hold off on closing that issue for the
moment, there's one other sub issue that I've just created a MR

that needs addressing before it can be closed.


"Daniel Sangorrin" <daniel.sangorrin@...> writes:

Thanks Robert, I will try today and see if it works.
# I'll close the gitlab issue in that case.

-----Original Message-----
From: cip-dev-bounces@... [mailto:cip-dev-bounces@...] On Behalf Of Robert Marshall
Sent: Wednesday, July 05, 2017 11:19 PM
To: cip dev
Subject: [cip-dev] running health checks from behind a web proxy

There is a known issue with b@d when running a lava QEMU health check
from behind a web proxy.

When attempting to retrieve the kernel, it produces the error

Invalid job data:
["HTTPSConnectionPool(host='', port=443):
Max retries exceeded with url: /kvm/standard/stretch-2.img.gz
(Caused by NewConnectionError('<requests.packages.urllib3.connection.VerifiedHTTPSConnection
object at 0x7f430c9b9c50>: Failed to establish a new connection: [Errno -5] No address associated with hostname',))"]


As b@d is intended to use locally built artefacts and the default QEMU health
check uses a remote kernel, a suggested work around for this test would be to copy
locally to /var/www/images/kernel-ci/qemu (a new directory)
and alter the health check to use the url

which should avoid the use of any web proxies.

You'll probably need a fairly clean copy of the VM with plenty of free
disk space to manage the extra copy of the kernel.

Thoughts? Hoping to discuss this further at tomorrow's open meeting.

cip-dev mailing list

Join { to automatically receive all group messages.