[
https://issues.apache.org/jira/browse/IGNITE-29305?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18146462#comment-18146462
]
Anton Vinogradov commented on IGNITE-29305:
-------------------------------------------
[Ignite PR Checker|https://ignite-pr-checker.is-a.dev/ignite?pr=13851] verdict
for PR 13851 · RunAll build
[9506352|https://ci2.ignite.apache.org/build/9506352] · tested
[ea299f5|https://github.com/apache/ignite/pull/13851/commits/ea299f569a6724f8d5388a815da76d26d6ef67c8]
· 149 suites ran, 0 reused
(!) *This run doesn't cover the PR fully:*
- 2 suites have no reliable result (2× execution timeout · non-zero exit code)
Everything below is what it did manage to say.
(!) *Broken suites* (no reliable result):
- execution timeout · non-zero exit code (2 suites): \[Check Code Style
Ducktests\], Thin client: Node.js
(?) *No blockers found — but the run above can't prove the PR is clean.* 110
failing tests are not blamed on this PR. Re-run after you handle what is listed
above.
_What each label means and what to do: [verdict
glossary|https://github.com/anton-vinogradov/ignite-pr-checker/blob/main/docs/features.md#verdict-glossary]._
♻️ Settled after 2 auto re-run waves: #1 — 7 suites that failed mid-run; #2 — 4
suites (1 with blockers, 3 broken).
> Start and stop the SSL factory registry with the node's component lifecycle
> ---------------------------------------------------------------------------
>
> Key: IGNITE-29305
> URL: https://issues.apache.org/jira/browse/IGNITE-29305
> Project: Ignite
> Issue Type: Sub-task
> Reporter: Anton Vinogradov
> Priority: Major
> Labels: ise
> Time Spent: 0.5h
> Remaining Estimate: 0h
>
> The SSL factory registry ({{ctx.ssl()}}) runs automatic TLS certificate hot
> reload: an hourly check thread and the {{ssl}} metrics. {{IgniteKernal}}
> starts it with a direct {{ctx.ssl().start(ctx)}} call after the components'
> {{onKernalStart}} and stops it with {{ctx.ssl().stop()}} at the beginning of
> {{stop0}}. Other node parts with a lifecycle are processors that the common
> component loop starts and stops, so the registry is the only one wired by
> hand.
> *What to do*
> * Make the registry a processor and start hot reload in {{onKernalStart}}.
> All four transports (discovery, communication, client connector, binary REST)
> register their factories during {{start()}}, so they are all registered by
> then.
> * Stop hot reload in {{onKernalStop}}, before the transports stop.
> * Remove the two direct calls from {{IgniteKernal}}.
> * Keep {{ctx.ssl()}} available in every context. {{GridKernalContextImpl}}
> creates the registry as a field, so SPIs that start in
> {{GridTestKernalContext}} without {{IgniteKernal}} can still register their
> factories. A processor created only in {{IgniteKernal}} would make
> {{ctx.ssl()}} null there.
> *Stop order.* The client connector and binary REST stop their servers in
> their own {{onKernalStop}}, and the node calls {{onKernalStop}} in reverse
> start order. Start the registry processor after {{ClientListenerProcessor}}
> and the REST processor, so that it stops hot reload before them.
> *Why a separate task*
> IGNITE-29211 gave the registry its lifecycle. Moving it to the common path
> changes how the context creates the registry and how test contexts get it,
> which is outside hot reload itself.
> *Done when*
> * {{IgniteKernal}} has no direct registry start or stop calls.
> * Hot reload starts after all transports registered their factories and stops
> before the transports stop.
> * The existing SSL hot reload and {{control.sh --ssl reload}} tests pass.
> *Order:* after IGNITE-29211, before IGNITE-29212.
--
This message was sent by Atlassian Jira
(v10.3.26#10030026)