Hi Or, This is a pretty scary report! If you haven't already, can you open a JIRA ticket describing how you detect and reproduce the problem? I'm reasonably familiar with the restore code, so I can help review if we do ultimately get a PR to fix this.you do ultimately put a PR together.
Best, Jason On Tue, Dec 24, 2024 at 12:22 PM Or Nagar <ornaga...@gmail.com> wrote: > > Hello, > > I’ve been working extensively with Apache Solr, specifically with > large-scale Solr Cloud collections, and I've encountered an issue that I’d > like to share with you. > > In Solr Cloud mode, when a collection has two or more replicas in a shard, > I noticed a significant problem when restoring from a backup to the shard's > leader. After the restore, the follower replica fails to start replication. > This is a critical issue, as it means that the Follower will not have the > updated data, rendering it Essentially useless. > > In the worst case, if the Leader shard becomes unavailable for any reason, > the outdated Follower could be promoted to leader. This would result in a > situation where the previously outdated Follower replica becomes the Leader > and the Leader may even start replicating from what used to be its own > replica - which will Cause a Full Loss of restoration data. > > I tested this behavior in a multi-sharded Solr collection with two replicas > per shard, and the issue was consistently reproducible. The restore is a > shard restore. > > I am more than willing to contribute to the necessary development to > resolve this if needed. --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@solr.apache.org For additional commands, e-mail: dev-h...@solr.apache.org
