Request timeout during index validation

Hibernate Search
21

This is… wild.

So between these two lines, it would seem that the whole application is stuck, doing absolutely nothing for 100s:

2022-07-26 11:36:48,221 o.hibernate.search.elasticsearch.request  [T] HSEARCH400093: Executed Elasticsearch HTTP GET request to 'http://127.0.0.1:9200' with path '/greffetribunal-write,greffetribunal-read' [...]
2022-07-26 11:38:28,162       o.h.s.e.r.spi.RootFailureCollector  [E] HSEARCH000521: Hibernate Search encountered a failure during bootstrap; continuing for now to list all problems, but the process will ultimately be aborted.

I’m trying to reproduce with a higher number of indexes/fields, but I’m having a hard time understanding what exactly is stuck…

22

Did you add the “wrong” mapping ? fullTextFieldSTANDARD

23

Yes.

I managed to reproduce the problem with a higher number of fields and errors. It seems caused by a deadlock in RootFailureCollector. So, I’m on it :slight_smile:

24

Nice. On my side, this only one (specific) error of validation is sufficient.

25

I’m guessing there’s only one cause, but the problem occurs on (almost) all you full-text fields, resulting in more than 100 failures in concurrent threads (roughly one thread per index), which triggers this: [HSEARCH-4654] - Hibernate JIRA

26

By failure do you mean mass indexation failures ? The MassIndexingFailureHandler is never called. All is ok.

27

No, I’m talking about validation on startup, which is where you problem is occurring. I’m talking about a validation failure, i.e. Hibernate Search detecting that a field in the Elasticsearch schema is not configured as expected, and thus validation failing. Though in your case that validation failure was caused by a bug ([HSEARCH-4652] - Hibernate JIRA)