Our migration application is built in java for migrating data into hyperledger fabric 2.2 using Java SDk to connect to HLF Network. We are processing data in batches of 4000 records in a scheduler and scheduler runs after every 1 min interval. We migrated 40,000 records in 10 minutes in after which, 993.689 Mb was occupied by Docker Containers (Peers, Orderers, CouchDB, CA, Chaincode). we stopped the java process and kept the system idle and after 1 hour it released around 200 Mb only.
We are trying to migrate 2.5 million records. After running the scheduler multiple times, system is getting stopped (no memory available). JVM is not leaking any memory, we have confirmed that. I want to understand why Docker containers are not releasing memory. Where is the memory leak happening?