[
https://issues.apache.org/jira/browse/SOLR-17719?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17938285#comment-17938285
]
Chris M. Hostetter commented on SOLR-17719:
-------------------------------------------
{quote} Wouldn't it be simpler to reach into the leader core to get an instance
of the schema (save a reference to this), then wait some time later for the
schema to be different?
{quote} # that would change the nature of the test
** i'm taking the test at face value that there is some reason why a specific
sequence of events (add field, add docs, wait for autoSoftCommit to make docs
searchable) is problematic, and we need to test that sequence and assert that
the docs are eventually searchable
** just testing that the index schema eventually changes doesn't ensure that
the docs are searchable
# i don't believe there is anyway to "await" for a new IndexSchema...
** I'm *NEVER* going to advocate for adding a sleep loop in a test (or
replacing a "sleep+assert" with a "loop(sleep+check); assert")
** awaiting on a queue (or a semaphore or some other inter-thread
communication) is much more reliable, takes up less cpu time, and takes up the
minimal about of wall clock necessary
> TestManagedSchemaWithMultipleAdd timing related failures (should use a
> semaphore hook to wait for newSearcher)
> --------------------------------------------------------------------------------------------------------------
>
> Key: SOLR-17719
> URL: https://issues.apache.org/jira/browse/SOLR-17719
> Project: Solr
> Issue Type: New Feature
> Reporter: Chris M. Hostetter
> Priority: Major
> Labels: newdev
>
> TestManagedSchemaWithMultipleAdd can fail sporadically – frequently on MacOS.
> The basic logic of the test is to:
> * add a field
> * immediately add some documents using that field
> * wait for autocommit and confirm we can search for those documents
> The problematic bit is the "wait for autocommit" – the test has the following
> bit of code...
> {noformat}
> // The issue we test in this class does not appear when using explicit
> commits.
> // Because of this we are waiting for autoSoftCommit to finish if there
> is one.
> Thread.sleep(AUTOSOFTCOMMIT_MAXTIME_MS + 500); assertEquals(
> numDocs, cloudClient.query(collection, new
> SolrQuery("*:*")).getResults().getNumFound());
> {noformat}
> ...but that assertion can fail anytime the autocommit timer sleep runs a
> little slower then the test sleep.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@solr.apache.org
For additional commands, e-mail: issues-h...@solr.apache.org
