I am facing lot of issues on latest production available couch-base server 5.0. Following are few:-
I upgraded it on DEV environment from 4.6.3. The older one was using high level of CPU even when the number of documents are less then half a million and only 1 sync gateway with max 50 mobile users and kafka connector. It got 9 buckets.
It got 100% CPU almost every time. There are 4 cores and 16 GB of RAM. swapiness and THP is configured as provided in the documents.
Even after upgrade the CPU is still 100% and Beam.smp , Indexer and Memchached using all the CPU.
Restarted the VM , added a node but then Rebalance is very slow even when 0.25 million documents will be transferred. it has already taken 2 hours in re-balance till the time of writing this.
queries are running slowly, the good part is the time out has been increased.
Memory foot print is low. But please help me reduce CPU usage and slowness.
Firstly, thank you so much for candid feedback on 5.0.
With every release especially the major ones performance is high focus for us. To get to the bottom of whats causing that slowness (since it involves both Server and SyncGateway) when you upgraded your DEV VM to 5.0 I would request you to file JIRA issue and add details about your DEV environment (Server & Sync Gateway versions), steps you performed in upgrading also run cbcollect_info to capture the logs which you can attach to issue for debugging purposes.
Secondly, appreciate your feedback feature functionality you like and what to improve.