[Cluster launcher] [Azure] add option for eviction policy in azure VM template#45397
Conversation
bthananjeyan
requested review from
a team,
architkulkarni,
ericl and
hongchaodeng
as code owners
bthananjeyan
force-pushed
the
brijen/delete-template
branch
from
7b8d4e3 to
312f792
Compare
|
@architkulkarni @ericl @hongchaodeng please let me know if you have any feedback on this PR, or if there is someone else who should be assigned to review it. Thanks! |
bthananjeyan
force-pushed
the
brijen/delete-template
branch
3 times, most recently
from
3967ace to
729bdca
Compare
anyscalesam
added
triage
Signed-off-by: bthananjeyan <brijen@ambirobotics.com>
bthananjeyan
force-pushed
the
brijen/delete-template
branch
from
729bdca to
0d8fc30
Compare
|
@architkulkarni @ericl @hongchaodeng bumping this PR in case it was buried. Please let me know if you have any feedback, or if someone else can be assigned to it. I don't have the ability to assign reviewers. |
… template (ray-project#45397) <!-- Thank you for your contribution! Please review https://github.com/ray-project/ray/blob/master/CONTRIBUTING.rst before opening a pull request. --> <!-- Please add a reviewer to the assignee section when you create a PR. If you don't have the access to it, we will shortly find a reviewer and assign them to your PR. --> ## Why are these changes needed? This is necessary to give users the option to pick an eviction policy (Delete vs Deallocate) when nodes in the cluster are pre-empted. <!-- Please give a short summary of the change and the problem this solves. --> This exposes an additional option that can be set in the `node_config` attribute in the cluster YAML file. ## Related issue number <!-- For example: "Closes ray-project#1234" --> ## Checks - [x] I've signed off every commit(by using the -s flag, i.e., `git commit -s`) in this PR. - [x] I've run `scripts/format.sh` to lint the changes in this PR. - [ ] I've included any doc changes needed for https://docs.ray.io/en/master/. - [ ] I've added any new APIs to the API Reference. For example, if I added a method in Tune, I've added it in `doc/source/tune/api/` under the corresponding `.rst` file. - [ ] I've made sure the tests are passing. Note that there might be a few flaky tests, see the recent failures at https://flakey-tests.ray.io/ - Testing Strategy - [ ] Unit tests - [ ] Release tests - [x] This PR is not tested :( --- - To see the specific tasks where the Asana app for GitHub is being used, see below: - https://app.asana.com/0/0/1207339070862431 Signed-off-by: bthananjeyan <brijen@ambirobotics.com> Signed-off-by: Richard Liu <ricliu@google.com>
|
FYI, I get an error:
which was patched for those users that do not deploy all spot instances: I got this error because my assumption is that head node is not a spot instance since you want this node to be stable. Is the best cost-effective practice to keep all nodes as spot instances (even the head node?) When I set all nodes (including the head) to be spot instances, ray boots up the cluster fine since the eviction policy is applicable to all the spot instances. Thanks |
|
The Pull request was closed because I hadn't tested it and when I came around to testing it, it turned out it didn't work. #46198 This was my original issue where I tracked the problem when it first arose. |
6 participants
Why are these changes needed?
This is necessary to give users the option to pick an eviction policy (Delete vs Deallocate) when nodes in the cluster are pre-empted.
This exposes an additional option that can be set in the
node_configattribute in the cluster YAML file.Related issue number
Checks
git commit -s) in this PR.scripts/format.shto lint the changes in this PR.method in Tune, I've added it in
doc/source/tune/api/under thecorresponding
.rstfile.