A bug to track this would be great. I've also noticed this behavior occur without nodes rejoining, so the root cause may not be due to that, but it probably more likely triggers the loop. I've tried the following configurations, and they all exhibit this issue:
- 1.9 with igfs-meta replicated - 2.0 with igfs-meta replicated - 2.0 with igfs-meta partitioned + 1 backup I don't think there's anything unusual with my configuration--my workload is a set of Spark jobs running at constant intervals. -- View this message in context: http://apache-ignite-users.70518.x6.nabble.com/igfs-meta-behavior-when-node-restarts-tp13155p13299.html Sent from the Apache Ignite Users mailing list archive at Nabble.com.