Ah, unless someone fixed it since the last time Mark Miller ranted about it, it's rooted in the fact that the call to create a collection returns to the user before the collection is ready to use. IIRC the rant said something like "what database would return from CREATE TABLE before the table was created? None! But we do just that with create Collection"... or something like that.
On Tue, Feb 6, 2024 at 4:56 PM David Smiley <dsmi...@apache.org> wrote: > Hey... about about that click-bait title; ehh? :-D > > I've noticed that basically any SolrCloudTestCase will create a > collection and then call waitForState. I think this is dumb; do we > ask our clients to do the same? Then why should our tests do this? I > can understand some tests may have specialized requirements -- fine. > But I think mostly we copy-paste this practice and not consider why > we're even doing this in the first place. This has been a code smell > to me. The only reason I can think of, and that which I actually > found in production today on a massive cluster, is that the > CloudSolrClient's state is out-of-date and CSC fails to consider that > it doesn't actually know the most up-to-date state, and fails a > request to send to the collection that it just created. > https://issues.apache.org/jira/browse/SOLR-17153 > Feedback welcome. > > ~ David Smiley > Apache Lucene/Solr Search Developer > http://www.linkedin.com/in/davidwsmiley > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@solr.apache.org > For additional commands, e-mail: dev-h...@solr.apache.org > > -- http://www.needhamsoftware.com (work) https://a.co/d/b2sZLD9 (my fantasy fiction book)
