3.5 Upstream pick list #13025
Comments
|
@pweil- I thought we didn't want the Docker security opt separator choosing logic as we didn't support the older Docker API version anyway |
|
@stevekuznetsov Not support doesn't means no one use it, just in case someone use it. If we have this, it will be better. I vote better contain this. |
|
@stevekuznetsov yeah, what @mdshuai mentioned is why I'm proposing we pick it. A) it went in to upstream with version specific support so we'll get it anyway in 1.6 and B) we're already receiving bugs on it being "broken". |
|
I thought @pmorie had done that for the explicit reason that we do not support that version of Docker and do not want to enable people to use it in the unsupported state. I'll let you guys duke it out :) |
|
then we need a min version check on startup RFE, not random code that fails 😄 |
|
@pweil- definitely want the AWS: recognize eu-west-2 region :-) |
|
@pweil- I am 100% certain that the installer will fail you if you try to install OCP on a system with an old Docker. |
|
I was able to install OCP 3.5 (with Docker 1.10.3) and installer did not fail, should I create a bug for that ? |
|
@sdodson seems like a bug ^^ |
I believe @sjenning / @derekwaynecarr reverted upstream changes that enabled "critical pods" in a |
|
Yeah, I think that particular instance was on atomic host where our only recourse is to abort the install/upgrade and have them upgrade to a newer version of atomic host. If for some reason they need downgrade, like 3.3 requiring docker-1.10 then they'll have to downgrade atomic host. |
|
We need:
As well. Both these PRs have been already backported to 3.4 . |
|
thanks @gnufied, added to the list. I see the 3.4 PRs but not master ones. Since you've already gone through the trouble of backporting to 3.4 can you open the master/3.5 PRs? If you want to respond on this thread or ping me the numbers in IRC I'll update the main comment. |
|
@pweil- Yes I will bring them to 3.5/master. I was going to in fact, just caught up with 1.6 freeze in upstream. |
|
@deads2k kubernetes/kubernetes@37cd179 - present request header cert CA (4 weeks ago) -> this does not apply cleanely, can you please check? |
|
@mfojtik looks like that's the backport we need to get rid of my patch. |
|
Also needed: kubernetes/kubernetes#42236 |
@pmorie since you're the original author of the PR do we still want to have it for 1.5? |
|
@pweil- and we are done :-) |
|
I think we're done with all cherry-picks, I'm closing this then. |
Suggested picks from the last 4 weeks in the kube 1.5 branch. @smarterclayton @ncdc @liggitt @mfojtik @deads2k PTAL and confirm/deny any that you can. Thanks.
AWS: Add sequential allocator for device names. kubernetes/kubernetes#38818 [1.5] Fix upstream AWS issues #13131Fix AWS device allocator to only use valid device names kubernetes/kubernetes#41455 [1.5] Fix upstream AWS issues #13131kubernetes/kubernetes@1bf90c8 - Change default reconciler sync period to 1 minute (7 days ago)(UPSTREAM: 41366: Change default reconciler sync period to 1 minute #13132)kubernetes/kubernetes@3151e6e - flag gate critical pods annotation support in kubeletDo we need the above? We never allowed those, did we?
kubernetes/kubernetes@81ac623 - cherry-pick #38855 - Fix variable shadowing in exponential backoff (13 days ago)UPSTREAM: 38855: Fix variable shadowing in exponential backoff when d… #13083 UPSTREAM: 38855: Fix variable shadowing in exponential backoff when d… #13084kubernetes/kubernetes@a80e219 - cherr-pick #38339 - Exponential back off when volume delete fails (13 days ago)done in 4615974kubernetes/kubernetes@3ccda07 - Plumb subresource through subjectaccessreview (2 weeks ago)UPSTREAM: 40935: Plumb subresource through subjectaccessreview #13085 UPSTREAM: 40935: Plumb subresource through subjectaccessreview #13086kubernetes/kubernetes@9903bd8 - Remove duplicate calls to describeInstance from aws (3 weeks ago)done in UPSTREAM: 39842: Remove duplicate calls to DescribeInstance #12740kubernetes/kubernetes@b1882b6 - controller: old pods should block deployment completeness (3 weeks ago)(UPSTREAM: 40625: controller: old pods should block deployment completeness #13133)kubernetes/kubernetes@e4f22b5 - Adjust global log limit to 1ms (4 weeks ago)UPSTREAM: 40497: Make HandleError prevent hot-loops #13087 UPSTREAM: 40497: Make HandleError prevent hot-loops #13088kubernetes/kubernetes@9fc7413 - Make HandleError prevent hot-loops (4 weeks ago)UPSTREAM: 40497: Make HandleError prevent hot-loops #13087 UPSTREAM: 40497: Make HandleError prevent hot-loops #13088kubernetes/kubernetes@37cd179 - present request header cert CA (4 weeks ago)UPSTREAM: 40301: present request header cert CA #13145kubernetes/kubernetes@7c187ca - Fixed bug which prevents PDBs which have minAvailable specified as a percentage working with StatefulSet pods (4 weeks ago)(UPSTREAM: 39825: Make PDBs represent percentage in StatefulSet #13143)kubernetes/kubernetes@2eef243 - various kubectl fixes (4 weeks ago)done in Kubernetes 1.5.0-beta.2 rebase #12143kubernetes/kubernetes@c529006 - Add path exist check in getPodVolumePathListFromDisk (4 weeks ago)UPSTREAM: 38909: Add path exist check in getPodVolumePathListFromDisk #13057, UPSTREAM: 38909: Add path exist check in getPodVolumePathListFromDisk #13058kubernetes/kubernetes@6941fe3 - Check if error is Status in result.Stream() (4 weeks ago)done in UPSTREAM: 39831: Check if error is Status in result.Stream() #12610kubernetes/kubernetes@ea100df - AWS: recognize eu-west-2 region UPSTREAM: 38746: recognize eu-west-2 region #13056,UPSTREAM: 38746: recognize eu-west-2 region #13055kubernetes/kubernetes@9eab226 - Enqueue controllers after minreadyseconds when all pods are ready UPSTREAM: 42097: Enqueue controllers after minreadyseconds when all pods are ready #13140Not from 1.5 branch:
docker opt separator (consider a backported version of this with docker version check rather than the existing fix)UPSTREAM: 40903: Set docker opt separator correctly for SELinux options #13141The text was updated successfully, but these errors were encountered: