Context
Observed on my local development cluster of 8-32 nodes.
When pull sync is ongoing (once every few minutes), it holds up retrieval / pushsync for 10-20 seconds.
Outside of that interval, the affected node serves requests, but while the pullsync logs/errors are printed in that 10-20 seconds nothing else is happening.
Steps to reproduce
Run a cluster of 10-30 nodes. Maintain retrieval / pushsync for a prolonged period of time by using a larger file. Observe refreshments / retrieves / pushsyncs in a single node logs for 5-10 minutes.
Context
Observed on my local development cluster of 8-32 nodes.
When pull sync is ongoing (once every few minutes), it holds up retrieval / pushsync for 10-20 seconds.
Outside of that interval, the affected node serves requests, but while the pullsync logs/errors are printed in that 10-20 seconds nothing else is happening.
Steps to reproduce
Run a cluster of 10-30 nodes. Maintain retrieval / pushsync for a prolonged period of time by using a larger file. Observe refreshments / retrieves / pushsyncs in a single node logs for 5-10 minutes.