Hello, I’m running an in-memory Ignite cluster in Kubernetes. For now it’s just two pods. I’m using it to cache the results of GRPC calls using readthrough mode and custom implementations of CacheLoader. I’m using drop wizard to start my grpc server and to start the Ignite server inside the same pod. There is a second remote grpc server that is the data store that ignite reads from on a cache miss.
In Partitioned mode, with zero backups, everything seems to work fine. I’ve been running it this way for a while. I’m anticipating an increase in activity soon, and so I wanted to do some load testing to compare Partitioned mode to Replicated mode for my expected load. But when I start it in Replicated mode, the two nodes don’t seem to agree with each other as changes are made to the cache. If I change a value on node zero and then query node one for that value, it will not have the changed value. I turned on debug logging and found it flooded with logs similar to below, and I’m getting those logs in both Partitioned and Replicated mode. 2025-04-09 12:45:13,085 DEBUG [grid-timeout-worker-#22] o.a.i.i.p.t.GridTimeoutProcessor - Timeout has occurred [obj=CancelableTask [id=39ec29a1691-2081261f-6109-4ea9-b675-64addc0f5001, endTime=1744202713076, period=1000, cancel=false, task=org.apache.ignite.internal.processors.cache.transactions.IgniteTxManager$$Lambda$1170/0x00007f80767d9b88@4bb534a0], process=true] 2025-04-09 12:45:13,539 DEBUG [grid-timeout-worker-#22] o.a.i.i.p.t.GridTimeoutProcessor - Timeout has occurred [obj=CancelableTask [id=d7ec29a1691-2081261f-6109-4ea9-b675-64addc0f5001, endTime=1744202713530, period=3000, cancel=false, task=org.apache.ignite.internal.processors.query.GridQueryProcessor$$Lambda$1097/0x00007f80767188e0@5d01cedc<mailto:task=org.apache.ignite.internal.processors.query.GridQueryProcessor$$Lambda$1097/0x00007f80767188e0@5d01cedc>], process=true] Are these logs anything to be concerned about? Are they pointing me at possible reasons that Replicated mode is not working? What are the possible reasons Replicated mode could fail to share updates to the cache? Thanks for any insights you can give me. Kurtis
