I did some data replication tests using V2.x (I first used V2.5 and upgraded to V2.7 later after I found the issues) between a local and a Microsoft Azure instant of ubuntu server. I could see the available memory dropped from 13G to 600M gradually in two weeks after I added the data replication to Azure server. When I restarted influxdb, the available memory reset to 13G. When running the “influx replication list ***”, I can see the Remaining Bytes to be Synced is always less than 10M for 8 buckets. Current Queue Bytes on Disk is always less than 100M for those buckets in total. The data replication itself works except that the constantly increasing memory usage could finally cause the system suspended.
Did I make some mistake to configure influxdb or the influxdb has some memory leakage issue for data replication feature?
Hello @tomzhang0910,
Welcome!
I don’t know if there’s a memory leak problem with data replication. In general operator control of memory is something that we’re looking to/working on giving 3.0 users. I don’t see any issues about memory leak with EDR on gh. But you might have better luck asking there.
However please know this feature isn’t receiving a lot of support unfortunately.
I am really appreciating your reply. I totally agree that it might not be an EDR issue. Maybe we are backing up too much data to Azure server using EDR. We will keep investigating on it.
