Hello!

I am running a large Solr 8.11 cluster in SolrCloud mode, and I've been
using the data import handler with a JDBC datasource triggered on a
schedule to keep my search index up to date. Now, as I look to upgrade the
cluster to >=9, I'm looking to migrate the job of the DIH to an external
scheduled batch job that reads the records from my datasource, transforms
them to Solr documents, and publishes them in batches to the "/update"
handler on my collection. I created a new collection and indexed it fully
using the new external batch process, however when I load test it, the Solr
processes on nodes that are hosting replicas of this new collection crash
with OOM exceptions under a fraction of the load (about 1/3) the
collection indexed with the DIH. Are there any performance concerns or
index construction concerns I should be aware of by flipping to the update
handler from the DIH? Let me know if you need any more information.

Thanks in advance,
Jack

Reply via email to