Populating the host_name field with cluster-local-ips. #1173
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
meta-cla
bot
added
the
CLA Signed
|
@johnwhumphreys has exported this pull request. If you are a Meta employee, you can view the originating Diff in D88567449. |
johnwhumphreys
added a commit
to johnwhumphreys/torchx
that referenced
this pull request
Dec 8, 2025
…1173) Summary: In the current implementation, the hostname is set to empty-string in the k8s scheduler's describe function (basically it's an implementation gap). This diff uses the k8s client to map the pod-name to its ip, and generate the pop cluster-dns <dashed-ip>.<namespace>.cluster.local and return it. If the pod IP is not available yet, this version of the diff will simply return an empty hostname since the k8s hostname is derived from the IP. The describe API is intended to quickly return the actual state from the scheduler's POV; so returning an empty host name when the IP is not yet assigned is better than blocking until it is, which could take an arbitrary amount of time. Existing users of the API should be fine as the hostname was not populuated anyway. New users will need to implement polling until the host name is ready, if they have not already. Differential Revision: D88567449
johnwhumphreys
added a commit
to johnwhumphreys/torchx
that referenced
this pull request
Dec 8, 2025
…1173) Summary: In the current implementation, the hostname is set to empty-string in the k8s scheduler's describe function (basically it's an implementation gap). This diff uses the k8s client to map the pod-name to its ip, and generate the pop cluster-dns <dashed-ip>.<namespace>.cluster.local and return it. If the pod IP is not available yet, this version of the diff will simply return an empty hostname since the k8s hostname is derived from the IP. The describe API is intended to quickly return the actual state from the scheduler's POV; so returning an empty host name when the IP is not yet assigned is better than blocking until it is, which could take an arbitrary amount of time. Existing users of the API should be fine as the hostname was not populuated anyway. New users will need to implement polling until the host name is ready, if they have not already. Differential Revision: D88567449
johnwhumphreys
force-pushed
the
export-D88567449
branch
from
3be7d89 to
6e88773
Compare
johnwhumphreys
added a commit
to johnwhumphreys/torchx
that referenced
this pull request
Dec 8, 2025
…1173) Summary: In the current implementation, the hostname is set to empty-string in the k8s scheduler's describe function (basically it's an implementation gap). This diff uses the k8s client to map the pod-name to its ip, and generate the pop cluster-dns <dashed-ip>.<namespace>.cluster.local and return it. If the pod IP is not available yet, this version of the diff will simply return an empty hostname since the k8s hostname is derived from the IP. The describe API is intended to quickly return the actual state from the scheduler's POV; so returning an empty host name when the IP is not yet assigned is better than blocking until it is, which could take an arbitrary amount of time. Existing users of the API should be fine as the hostname was not populuated anyway. New users will need to implement polling until the host name is ready, if they have not already. Differential Revision: D88567449
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #1173 +/- ##
==========================================
+ Coverage 91.70% 91.71% +0.01%
==========================================
Files 86 86
Lines 6653 6665 +12
==========================================
+ Hits 6101 6113 +12
Misses 552 552
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
johnwhumphreys
added a commit
to johnwhumphreys/torchx
that referenced
this pull request
Dec 8, 2025
…1173) Summary: In the current implementation, the hostname is set to empty-string in the k8s scheduler's describe function (basically it's an implementation gap). This diff uses the k8s client to map the pod-name to its ip, and generate the pop cluster-dns <dashed-ip>.<namespace>.cluster.local and return it. If the pod IP is not available yet, this version of the diff will simply return an empty hostname since the k8s hostname is derived from the IP. The describe API is intended to quickly return the actual state from the scheduler's POV; so returning an empty host name when the IP is not yet assigned is better than blocking until it is, which could take an arbitrary amount of time. Existing users of the API should be fine as the hostname was not populuated anyway. New users will need to implement polling until the host name is ready, if they have not already. Reviewed By: AbishekS Differential Revision: D88567449
johnwhumphreys
force-pushed
the
export-D88567449
branch
from
6e88773 to
b04d132
Compare
johnwhumphreys
added a commit
to johnwhumphreys/torchx
that referenced
this pull request
Dec 8, 2025
…1173) Summary: In the current implementation, the hostname is set to empty-string in the k8s scheduler's describe function (basically it's an implementation gap). This diff uses the k8s client to map the pod-name to its ip, and generate the pop cluster-dns <dashed-ip>.<namespace>.cluster.local and return it. If the pod IP is not available yet, this version of the diff will simply return an empty hostname since the k8s hostname is derived from the IP. The describe API is intended to quickly return the actual state from the scheduler's POV; so returning an empty host name when the IP is not yet assigned is better than blocking until it is, which could take an arbitrary amount of time. Existing users of the API should be fine as the hostname was not populuated anyway. New users will need to implement polling until the host name is ready, if they have not already. Reviewed By: AbishekS Differential Revision: D88567449
johnwhumphreys
added a commit
to johnwhumphreys/torchx
that referenced
this pull request
Dec 8, 2025
…1173) Summary: In the current implementation, the hostname is set to empty-string in the k8s scheduler's describe function (basically it's an implementation gap). This diff uses the k8s client to map the pod-name to its ip, and generate the pop cluster-dns <dashed-ip>.<namespace>.cluster.local and return it. If the pod IP is not available yet, this version of the diff will simply return an empty hostname since the k8s hostname is derived from the IP. The describe API is intended to quickly return the actual state from the scheduler's POV; so returning an empty host name when the IP is not yet assigned is better than blocking until it is, which could take an arbitrary amount of time. Existing users of the API should be fine as the hostname was not populuated anyway. New users will need to implement polling until the host name is ready, if they have not already. Reviewed By: AbishekS Differential Revision: D88567449
johnwhumphreys
force-pushed
the
export-D88567449
branch
from
b04d132 to
17e1ac7
Compare
johnwhumphreys
added a commit
to johnwhumphreys/torchx
that referenced
this pull request
Dec 8, 2025
…1173) Summary: In the current implementation, the hostname is set to empty-string in the k8s scheduler's describe function (basically it's an implementation gap). This diff uses the k8s client to map the pod-name to its ip, and generate the pop cluster-dns <dashed-ip>.<namespace>.cluster.local and return it. If the pod IP is not available yet, this version of the diff will simply return an empty hostname since the k8s hostname is derived from the IP. The describe API is intended to quickly return the actual state from the scheduler's POV; so returning an empty host name when the IP is not yet assigned is better than blocking until it is, which could take an arbitrary amount of time. Existing users of the API should be fine as the hostname was not populuated anyway. New users will need to implement polling until the host name is ready, if they have not already. Reviewed By: AbishekS Differential Revision: D88567449
…1173) Summary: In the current implementation, the hostname is set to empty-string in the k8s scheduler's describe function (basically it's an implementation gap). This diff uses the k8s client to map the pod-name to its ip, and generate the pop cluster-dns <dashed-ip>.<namespace>.cluster.local and return it. If the pod IP is not available yet, this version of the diff will simply return an empty hostname since the k8s hostname is derived from the IP. The describe API is intended to quickly return the actual state from the scheduler's POV; so returning an empty host name when the IP is not yet assigned is better than blocking until it is, which could take an arbitrary amount of time. Existing users of the API should be fine as the hostname was not populuated anyway. New users will need to implement polling until the host name is ready, if they have not already. Note: adding extra test and resubmitting to avoid GitHub coverage block. Reviewed By: AbishekS Differential Revision: D88567449
johnwhumphreys
force-pushed
the
export-D88567449
branch
from
17e1ac7 to
092156f
Compare
johnwhumphreys
added a commit
to johnwhumphreys/torchx
that referenced
this pull request
Dec 9, 2025
…1173) Summary: In the current implementation, the hostname is set to empty-string in the k8s scheduler's describe function (basically it's an implementation gap). This diff uses the k8s client to map the pod-name to its ip, and generate the pop cluster-dns <dashed-ip>.<namespace>.cluster.local and return it. If the pod IP is not available yet, this version of the diff will simply return an empty hostname since the k8s hostname is derived from the IP. The describe API is intended to quickly return the actual state from the scheduler's POV; so returning an empty host name when the IP is not yet assigned is better than blocking until it is, which could take an arbitrary amount of time. Existing users of the API should be fine as the hostname was not populuated anyway. New users will need to implement polling until the host name is ready, if they have not already. Note: adding extra test and resubmitting to avoid GitHub coverage block. Reviewed By: AbishekS Differential Revision: D88567449
johnwhumphreys
added a commit
to johnwhumphreys/torchx
that referenced
this pull request
Dec 9, 2025
…1173) Summary: In the current implementation, the hostname is set to empty-string in the k8s scheduler's describe function (basically it's an implementation gap). This diff uses the k8s client to map the pod-name to its ip, and generate the pop cluster-dns <dashed-ip>.<namespace>.cluster.local and return it. If the pod IP is not available yet, this version of the diff will simply return an empty hostname since the k8s hostname is derived from the IP. The describe API is intended to quickly return the actual state from the scheduler's POV; so returning an empty host name when the IP is not yet assigned is better than blocking until it is, which could take an arbitrary amount of time. Existing users of the API should be fine as the hostname was not populuated anyway. New users will need to implement polling until the host name is ready, if they have not already. Note: adding extra test and resubmitting to avoid GitHub coverage block. Reviewed By: AbishekS Differential Revision: D88567449
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Summary:
In the current implementation, the hostname is set to empty-string in the k8s scheduler's describe function. This diff uses the k8s client to map the pod-name to its ip, and generate the pop cluster-dns ..cluster.local and return it. If it fails, it makes a best attempt at a fallback.
Note: The fallback does not match prior functionality; but diff #4 in this stack makes it fall back to empty-string and adds tests to cover this diff.
Differential Revision: D88567449