We have a 1.7.1 cluster that I just setup a new repo on for snapshots, so I had to restart the cluster.
Was a graceful restart. Cluster hadn't been restarted in quite a while (months).
When it came up we had a number of unassigned replicas. I manually rerouted these, but now I have a number of replica shards that are stuck on initializing.
Some of these are pretty big (15-30gb) due to some poor choices on num of shards (we'll be correcting that after we move to 2.3.2).
Is there a way to specifically delete and recreate a specific replica? Or do I have to set the number of replicas to 0 then back up to 1 to recreate all replicas?
That's not that big really.[quote="ronchalant, post:1, topic:49459"]
Is there a way to specifically delete and recreate a specific replica? Or do I have to set the number of replicas to 0 then back up to 1 to recreate all replicas?
[/quote]
yeah by big I meant I think shards of 15-30gb for a growing index is going on the large side? I recently created a new index that we're expecting to grow quickly - it'll start at around 2TB and will grow about 1TB/year. Not super huge by Elastic standards, but it has me looking much more closely at shard allocation than I had previously.
Regarding the issue here in the thread, I had to step away from this for a while and the shards did eventually initialize after a couple hours. It just didn't make much sense to me since these indexes in question are relatively small overall.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.