NextbrickInc commented on code in PR #4818:
URL: https://github.com/apache/solr/pull/4818#discussion_r3870950968
##########
solr/webapp/src/test/org/apache/solr/webapp/AdminUiSchemaDesignerTest.java:
##########
@@ -25,11 +24,9 @@
* Happy-path test of the Schema Designer screen: create a new schema, paste a
sample document and
* let the designer analyze it.
*
- * <p>AwaitsFix: the designer backend transiently fails its own prep/analyze
calls ("version
- * mismatch, retry", "Error loading solr config") when driven at automation
speed, making this test
- * flaky even with retries.
+ * <p>The Analyze action remains disabled until creation of the mutable schema
has completed, so a
+ * fast user cannot race the prep and analyze requests.
*/
[email protected](bugUrl =
"https://issues.apache.org/jira/browse/SOLR-18347";)
Review Comment:
Thanks @epugh — two things that might change the calculus.
**It isn't flaky any more.** The `@AwaitsFix` blamed "transient failures at
automation speed"; the
real cause was deterministic. The designer addressed Luke by **core** name,
and a core name only
resolves on the node hosting it — with `configureCluster(2)`, that's the
wrong node about half the
time. Addressing the **collection** fixes it: 53 tests, 18 suites, 0
failures.
**Your hunch is right.** `SchemaDesignerResponse.core` came in with
SOLR-18152 (#4203), commented
*"Active replica core name ... used for Luke API calls"* —
reasonable-looking, just not routable
from an arbitrary node. Nothing else in that migration looks affected; every
Schema Designer
assertion passes on top of it.
So I'd keep the test rather than drop the page — it's what would have caught
this. And the V2
client would issue the same request against the same core name, so this
class of bug survives the
migration either way.
---
**Status — I think this is ready to go in.**
- @janhoy's three comments are answered; the one that needed a change is
pushed as `9b5545ddb4`.
- Full Admin UI Selenium suite: 53 tests, 18 suites, 0 failures, 0 skipped.
`tidy` clean,
`PingRequestHandlerTest` 6/6.
- No open review threads and nothing outstanding on my side.
@janhoy — the only reservation you raised was the five-in-one packaging. If
you're willing to take
it as-is, I'd appreciate an approval. If not, say so and I'll split it along
`PingRequestHandler` + Core Overview vs. the four UI-only fixes today.
@epugh — if the Schema Designer question above is settled for you, a second
approval would let this
land.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
