folks! Hope you're doing well. I have a solr cluster in cloud configured with 1 zk and 2 solr nodes. With a collection that has 1 shard and 2 replicas (one per node) configured as NRT. When i connect to zk and start to send updates (usually in batches of 100-1000 docs), the update request stalls and throws an IOException signaling that: "request processing has stalled for with remaining elements in the queue". But the docs barely have 2-3kb of average size. I'd like to have a clue about what i have to investigate. I want to know if it can be a slow network or something intrinsic to solr that i have to investigate. Like tweak the config files to see what happens. I can't share the solr config files, unfortunately. But one interesting thing is that the commits are done by the clients(usualy soft commits), there's no autoCommit configured. I know there are a few related questions, but the only answer i got is to increase the threshold limit for the request to stale. But this only masks the problem, not solve him.
I appreciate any help, Best Regards! :)
Already raised the stall threshold limit, but it only masks the slowness problem.
UPDATE: We discovered the problem. It was a field used for spatial search that holds sometimes very large polygons and lines. So we had to compress it before the index updates. Thank you.