Search index rebuild repeatedly returns connection refused multiple hours into operation


We've been having a lot of trouble with elasticsearch lately, which I had chalked up to it being out of date. Search index rebuilds kept failing, usually with a curl failure due to a refused connection (which makes no sense, because literally every single time I received a curl failure, I was able to manually curl the elasticsearch instance from the machine running our forums). I had hoped updating our machine running elasticsearch (as well as elasticsearch itself) would help resolve the problem, but the rebuild continues to give me a curl failure, and now I appear to be right back at square one. What am I supposed to do to rebuild search indexes for a forum that has 15+ years of data? Is there an incremental rebuild option? I've tried doing it via CLI and via the admin panel, and even resuming the operation from the admin panel will eventually cause it to fail.

Elasticsearch for sure is not running out of space, because the data directory is sitting on a dedicated attached disk with 100GB of space (it's using about 3GB now). Is is throttling connections or something? The admin panel tells me elasticsearch has yellow status, but again, I can curl it from our webserver with no issues. I also checked the elasticsearch log files to see if there were any errors showing up, but there's none whatsoever after I fixed a disk space issue. I've been messing with this for days, now, so I would really appreciate it if someone can point me in the right direction.
Top Bottom