Consolidate and bump default AWS instance type (m6i.xlarge)#2076
Conversation
|
/assign @jstuever |
openshift-ci
Bot
added
the
approved
2uasimojo
force-pushed
the
HIVE-2235/aws-instance-type
branch
from
2af43fe to
e8e40f8
Compare
Codecov Report
Additional details and impacted files@@ Coverage Diff @@
## master #2076 +/- ##
==========================================
+ Coverage 57.37% 57.39% +0.01%
==========================================
Files 186 186
Lines 25712 25713 +1
==========================================
+ Hits 14752 14757 +5
+ Misses 9719 9717 -2
+ Partials 1241 1239 -2
|
|
/test e2e-pool This didn’t even get around to creating clusters. And hive-operator was in a weird hot loop. None of which should be related to an instance type change. |
|
Oh. No, that's actually broken. I think I missed a path through clusterpools somehow. |
2uasimojo
force-pushed
the
HIVE-2235/aws-instance-type
branch
from
e8e40f8 to
e2d270d
Compare
This was just the namespace watch filter. We weren't actually reconciling (despite the message). |
2uasimojo
force-pushed
the
HIVE-2235/aws-instance-type
branch
from
e2d270d to
bbe6e70
Compare
|
/test e2e unrelated flake |
This commit cleans up how we handle instance types in hiveutil and e2e CI. - `hiveutil create-cluster` now accepts an `--aws-instance-type` option. - The default instance type is now `m6i.xlarge`. It was previously `m5.xlarge` for hiveutil and `m4.large` for the MachinePool scaling tests in our e2e CI. - This default now lives in one place, so it's easier to change later without missing things. - To allow easier noodling via CI rehearsals, `e2e` will feed a new `$AWS_INSTANCE_TYPE` environment variable through to the new `--aws-instance-type` param for `create-cluster`. HIVE-2235
2uasimojo
force-pushed
the
HIVE-2235/aws-instance-type
branch
from
bbe6e70 to
8574253
Compare
|
/test coverage known taint flake /test e2e-pool I have absolutely no idea what happened here. It looks like we got past the previous error, but then something weird happened with the uninstall of the ("real") pool cluster. |
|
Looks like 4.14 starts to use |
This sounds like a really useful tip... if only I had some context. Also, if this were a problem, wouldn't e2e have failed as well? It's using the same install-config builder as e2e-pool. |
|
This actually looks like it pretty much succeeded, but ran up against the overall 2h test timeout. /test e2e-pool Even if this eventually works, we're obviously skating way too close to the maximum allowable test time. I'm not even sure what we would do about that. |
|
I'm only guessing the slowdown has something to do with having increased the instance size. (Bigger instances take longer to... provision? hibernate? resume?) I believe I found about half an hour we can trim out of the test via #2080 -- will retest when that lands. |
|
Sorry there might be a mix-up, I wan't referring to the e2e-pool job fail. What I meant is, in addition to the default AWS instance type (generated by hiveutil), do you think it's a good idea to also update the networkType (to align with Openshift Installer's default) ? |
|
/test e2e-pool #2080 has landed (and it does seem to have had the desired effect). |
|
/lgtm |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: 2uasimojo, jstuever The full list of commands accepted by this bot can be found here. The pull request process is described here DetailsNeeds approval from an approver in each of these files:
Approvers can indicate their approval by writing |
Gosh, I don't know. Is there a benefit to doing so? Will it make something faster or cheaper? Is OpenShiftSDN being deprecated/removed? |
|
@2uasimojo: all tests passed! Full PR test history. Your PR dashboard. DetailsInstructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. I understand the commands that are listed here. |
I'd be curious if we could remove NetworkType and let things naturally use their defaults. |
Okay, sure: #2085 Note: I'm pretty sure this is only affecting hiveutil (and thereby e2e). |
4 participants
This commit cleans up how we handle instance types in hiveutil and e2e CI.
hiveutil create-clusternow accepts an--aws-instance-typeoption.m6i.xlarge. It was previouslym5.xlargefor hiveutil andm4.largefor the MachinePool scaling tests in our e2e CI.e2ewill feed a new$AWS_INSTANCE_TYPEenvironment variable through to the new--aws-instance-typeparam forcreate-cluster.HIVE-2235