Project

General

Profile

Actions

Bug #27048

open

s3aws: RequestError: send request failed after upgrade to mimic

Added by Dieter Roels over 5 years ago. Updated over 5 years ago.

Status:
New
Priority:
Normal
Assignee:
-
Target version:
-
% Done:

0%

Source:
Community (user)
Tags:
Backport:
Regression:
No
Severity:
3 - minor
Reviewed:
Affected Versions:
ceph-qa-suite:
Pull request ID:
Crash signature (v1):
Crash signature (v2):

Description

We have a multisite setup running luminous which is primarily used as the storage for our docker registry. Recently we upgraded a test cluster to mimic. The upgrade was successful, everything is healthy, but the docker registry no longer works. With s3cmd everything is ok, but when the docker registry tries to access its s3 backend it gives a unexpected EOF:

time="2018-08-22T09:24:27.900996761Z" level=error msg="response completed with error" auth.user.name=admin err.code="manifest unknown" err.detail="s3aws: RequestError: send request failed\ncaused by: Get https://<loadbalancerIP>/tda-registry/docker/registry/v2/repositories/pnu/kubectl-download/_manifests/tags/v1.11.2/current/link: EOF" err.message="manifest unknown" go.version=go1.7.6

We see the request on the loadbalancer, we see the request coming in on the radosgw server itself with tcpdump, but we do not see the request at all in radosgw logs, even with "debug rgw = 20". We are running civetweb with ssl, so we cannot see what civetweb/radosgw is returning from the tcpdumps, but when we run strace on the radosgw process we see the connection from the loadbalancer. So radosgw receives the request, but logs nothing.

Any tips on how we can debug this issue further?

Actions

Also available in: Atom PDF