Skip to content

Add opt-in isolated SolrCloud verification for SPLITSHARD and MIGRATE #232

Open

Description

@vsimko

Context

SPLITSHARD and MIGRATE have typed V1 parameter/response contracts and serialization/type fixtures, but no runtime verification. The coverage matrix intentionally excludes them from shared CI because they are long-running and can affect routing or data placement.

Scope

  • Design an opt-in isolated SolrCloud test environment with disposable collections and controlled compositeId routing.
  • Verify SPLITSHARD and MIGRATE semantics, completion/status behavior, and resulting shard/data placement.
  • Ensure cleanup and failure recovery cannot affect shared fixtures or external Solr clusters.
  • If safe automation is impractical, retain explicit non-certification and document a reproducible manual verification procedure.

Acceptance criteria

  • Tests are opt-in and cannot run against the default/shared integration collection.
  • Setup is guarded by explicit opt-in, loopback-only endpoint, SolrCloud, and multi-node requirements; fixtures create unique disposable collections, require a one-shard split source, and require an empty migration target.
  • Behavior is asserted from resulting cluster state and data placement (active child shards and split read-back; migrated target document read-back).
  • Timeout and failure handling are documented; async status is polled with a bounded timeout, status records are deleted on completion, and a failed cleanup requires discarding the ephemeral cluster. Live cancellation/partial-operation recovery are explicitly not certified.
  • Compatibility status changes only for versions actually tested.

Implementation update (2026-11-04)

  • PR test(solr): verify replication and shard lifecycle operations #253 adds an opt-in, loopback-only two-node SolrCloud workflow for SPLITSHARD and MIGRATE, using unique disposable collections, asynchronous status polling, post-operation state/read-back assertions, and cleanup.
  • The workflow is wired to workflow_dispatch and the enable_destructive_solrcloud_tests input. It is deliberately not run in ordinary PR CI.
  • The opt-in workflow still needs a manual dispatch and successful per-version evidence. Until then, compatibility remains unverified and this issue stays open.

Implementation status after PR #253 merge (2026-11-04)

  • PR test(solr): verify replication and shard lifecycle operations #253 is merged. The opt-in test harness and workflow are now on master.
  • Four safety/behavior/documentation acceptance criteria are implemented. The remaining gate is execution of the manual workflow_dispatch with enable_destructive_solrcloud_tests=true and successful per-version evidence for Solr 8.11.4, 9.11.0, and 10.1.0.
  • The available GitHub integration does not expose a workflow-dispatch action, so no run has been started from this session. Compatibility remains unverified until a repository maintainer manually runs the workflow and records the result.

Activity

  1. vsimko commented on Nov 4, 2026

    @vsimko
    OwnerAuthor

    Progress update (2026-11-04): PR #253 adds the opt-in, loopback-only, two-node SolrCloud workflow and isolated SPLITSHARD/MIGRATE assertions, with timeout/cleanup guidance. These destructive tests are intentionally skipped in ordinary CI; acceptance still requires a manual workflow_dispatch run with enable_destructive_solrcloud_tests=true on Solr 8.11.4, 9.11.0, and 10.1.0. The available GitHub integration here can inspect runs but cannot dispatch workflows, so leave this issue open until those manual runs are completed and their evidence is recorded.

  2. vsimko commented on Nov 4, 2026

    @vsimko
    OwnerAuthor

    Update (2026-11-04): PR #254 has been merged to master, so the CI workflow now exposes workflow_dispatch and the enable_destructive_solrcloud_tests input (default false). Manual SPLITSHARD/MIGRATE verification can now be started from Actions → CI → Run workflow, selecting test/solr-nonblocking-issues-final and setting the input to true. The opt-in run still needs to complete successfully for each pinned Solr version before this issue can be closed. PR #253 CI run #635 is currently in progress; its replication lifecycle matrix has not yet fully completed.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions