Repository navigation
Conversation
github-project-automation
Bot
moved this to Needs Triage
in Nomad - Community Issues Triage
mismithhisler
previously requested changes
Jun 15, 2026
mismithhisler
left a comment
mismithhisler
left a comment
Member
There was a problem hiding this comment.
Hi @eyupcanakman, thanks for the contribution! The detailed description helped provide good background here for my review. Overall looks good with one suggestion.
| // sticky is only supported on dynamic host volumes. Reject it on any static | ||
| // host volume up front, before the per-volume checks below, which can return | ||
| // early for another volume in the same request set. | ||
| for _, req := range h.volumeReqs { |
Member
There was a problem hiding this comment.
Since we're already iterating the volumeReqs below, can we move some of this logic there?
Contributor
Author
There was a problem hiding this comment.
Done, moved it into the static-volume branch of the loop. I switched the two sticky checks from early returns to continue so the loop still reaches a later volume, otherwise a sticky dynamic volume ahead of the static one in volumeReqs could short-circuit the check.
eyupcanakman
force-pushed
the
fix/feasibility-sticky-static-host-volume
branch
from
June 16, 2026 16:03
5d38813 to
869e782
Compare
tgross
self-requested a review
June 29, 2026 18:57
tgross
requested changes
Jun 29, 2026
tgross
left a comment
tgross
left a comment
Member
There was a problem hiding this comment.
HI @eyupcanakman! Sorry about the delayed re-review. I'm working on a bug in this same neck of the woods, so if you're not going to be able to get back to these remarks let me know and I can carry the PR.
sticky is only supported on dynamic host volumes. A sticky request that
resolved to a static host volume passed feasibility checking, so the alloc
was placed and plan apply then failed inserting a claim keyed by the empty
volume ID ("object missing primary index").
Check each host volume request in its own pass so the node is rejected when
any request is infeasible. This also fixes a case where a satisfied sticky
request short-circuited the checks for the other requests in the same group.
eyupcanakman
force-pushed
the
fix/feasibility-sticky-static-host-volume
branch
from
June 30, 2026 21:17
869e782 to
682fa5c
Compare
Contributor
Author
|
Got back to all of these. Reworked the loop into a per-request check, added a changelog entry for the second bug, and made the two tests deterministic. |
tgross
added
backport/ent/1.10.x+ent
github-project-automation
Bot
moved this from In Progress
to Done
in Nomad - Community Issues Triage
2 of 3 tasks
mismithhisler
pushed a commit
that referenced
this pull request
Jul 15, 2026
sticky is only supported on dynamic host volumes. A sticky request that
resolved to a static host volume passed feasibility checking, so the alloc
was placed and plan apply then failed inserting a claim keyed by the empty
volume ID ("object missing primary index").
Check each host volume request in its own pass so the node is rejected when
any request is infeasible. This also fixes a case where a satisfied sticky
request short-circuited the checks for the other requests in the same group.
schmichael
pushed a commit
to vercel/nomad
that referenced
this pull request
Sep 25, 2026
sticky is only supported on dynamic host volumes. A sticky request that
resolved to a static host volume passed feasibility checking, so the alloc
was placed and plan apply then failed inserting a claim keyed by the empty
volume ID ("object missing primary index").
Check each host volume request in its own pass so the node is rejected when
any request is infeasible. This also fixes a case where a satisfied sticky
request short-circuited the checks for the other requests in the same group.
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
Description
Setting
sticky = trueon a static host volume passed feasibility checking, so the scheduler placed the allocation and then failed to apply the plan withTask group volume claim insert failed: object missing primary index. Static host volumes have an empty ID, and the sticky claim is keyed by volume ID, so the claim insert hits an empty primary index. Static and dynamic host volumes share the sametype = "host"jobspec, so the distinction is only known once a node is selected, which is why this can't be caught at job submission.The fix rejects
stickyon a static host volume during feasibility checking, before the allocation is placed. The check runs before the per-volume loop, so a feasible dynamic sticky volume in the same task group can't return early and skip it. A job that setssticky = trueon a static host volume will now fail feasibility with a clear constraint instead of failing the evaluation. Removing thestickyflag restores placement.Testing & Reproduction steps
Added two tests.
TestHostVolumeChecker_StaticStickycovers the feasibility checker directly. Sticky read-write and read-only on a static volume are infeasible, and non-sticky stays feasible.TestServiceSched_JobRegister_StickyStaticHostVolumeruns the service scheduler end to end. Onmainit fails with theobject missing primary indexclaim error, and with the fix the evaluation completes with the allocation left unplaced and the host volume constraint recorded.go test ./scheduler/feasible/ ./scheduler/passes.Links
Fixes #27153
Per @tgross's note on the issue, this can't be caught at job submission because the static/dynamic distinction isn't known until feasibility checking.
Contributor Checklist
.changelog/27153.txt.Changes to Security Controls
No changes to security controls.
This contribution was developed with AI assistance (Claude Code), used to trace the feasibility path and draft the fix and tests.