tree aed065a2bc3368a55a29ece700215dff02cf66c8
parent 18d3169a368d0a55a941a7f6265829bd27e2eacb
author James E. Blair <jeblair@redhat.com> 1504888477 -0700
committer James E. Blair <jeblair@redhat.com> 1504888676 -0700

Revert "Enable maintainConnectionCache"

This reverts commit 517fd7741d1a1b56ab710ca3a7a1bdb2e836cc40.

2017-09-08 00:12:07,550 ERROR zuul.Scheduler: Exception in management event:
Traceback (most recent call last):
  File "/usr/local/lib/python3.5/dist-packages/zuul/scheduler.py", line 783, in process_management_queue
    self._doTenantReconfigureEvent(event)
  File "/usr/local/lib/python3.5/dist-packages/zuul/scheduler.py", line 485, in _doTenantReconfigureEvent
    self._reconfigureTenant(tenant)
  File "/usr/local/lib/python3.5/dist-packages/zuul/scheduler.py", line 588, in _reconfigureTenant
    self.maintainConnectionCache()
  File "/usr/local/lib/python3.5/dist-packages/zuul/scheduler.py", line 731, in maintainConnectionCache
    connection.maintainCache(relevant)
  File "/usr/local/lib/python3.5/dist-packages/zuul/driver/github/githubconnection.py", line 574, in maintainCache
    for key, change in self._change_cache.items():

This causes tenant reconfiguration to abort halfway through which can
cause changes to get stuck in a gate pipeline.

Specifically to reproduce this error, we need two changes in gate, the first
is a config (zuul.yaml) change, and some pull requests in the github cache.

When the first change lands, zuul will perform a reconfiguration and fail.
Then when the second change completes jobs, it will report and claim to be
removed from the queue, but it will still actually be there.

Change-Id: I32e4a4f864ad16be391f57793f1a3abecdd6eee8