Wait on free ports on GKE Autopilot #2901
Conversation
zmerlynn
added
the
feature-freeze-do-not-merge
|
Build Succeeded 👏 Build Id: 8024765f-e26c-4424-aa00-579d708bcfaf The following development artifacts have been built, and will exist for the next 30 days:
A preview of the website (the last 30 builds are retained): To install this version:
|
zmerlynn
force-pushed
the
wait-on-free-ports
branch
from
e191d0c to
1993699
Compare
|
Build Succeeded 👏 Build Id: 609bb904-d13b-4587-a235-2b6bc4070d3c The following development artifacts have been built, and will exist for the next 30 days:
A preview of the website (the last 30 builds are retained): To install this version:
|
|
Build Succeeded 👏 Build Id: 0741c7f1-1ce6-472f-a41a-f8a852d0e720 The following development artifacts have been built, and will exist for the next 30 days:
A preview of the website (the last 30 builds are retained): To install this version:
|
Kalaiselvi84
removed
the
feature-freeze-do-not-merge
zmerlynn
force-pushed
the
wait-on-free-ports
branch
from
653b45e to
7c77534
Compare
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: gongmax, markmandel, zmerlynn The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
Build Failed 😱 Build Id: ed5b5423-d79e-4c0f-a8ee-30040bf93bd7 To get permission to view the Cloud Build view, join the agones-discuss Google Group. |
On Autopilot, since port assignments are random, port conflicts happen. Just accept it and wait on the autoscaler to kick in - the problem will go away. We do this by introducing a WaitOnFreePorts() cloud product hook to indicate whether we should pay attention to the case where the pod is unschedulable due to free ports. It is very likely this condition occurs organically in any autoscaling scenario, but the fleet controller also hides the condition by rapidly deleting the Unhealthy GameServer. Along the way: * fix a typo in the e2e framework that I introduced * change TestGameServerRestartBeforeReadyCrash to wait for the framework default - moving to `Scheduled` can be long (e.g. this exact PR)
zmerlynn
force-pushed
the
wait-on-free-ports
branch
from
7c77534 to
7f9bce9
Compare
|
New changes are detected. LGTM label has been removed. |
|
Build Succeeded 👏 Build Id: 82cbea86-1c4e-48b5-969a-4b93c4d6fcc9 The following development artifacts have been built, and will exist for the next 30 days:
A preview of the website (the last 30 builds are retained): To install this version:
|
On Autopilot, since port assignments are random, port conflicts happen. Just accept it and wait on the autoscaler to kick in - the problem will go away.
We do this by introducing a
WaitOnFreePorts()cloud product hook to indicate whether we should pay attention to the case where the pod is unschedulable due to free ports.It is very likely this condition occurs organically in any autoscaling scenario, but the fleet controller also hides the condition by rapidly deleting the Unhealthy GameServer.
Along the way, fix a typo I introduced in the e2e framework.