when server one node down(not use failover),I get the error code LCB_CNTL_DETAILED_ERRCODES,the memory leak,How can solution？
Can you explain exactly what leaks memory?
- when couchbase server one node down,the client request(10000 ops/s) the couchbase server cluster,the c sdk memory always increase.the server status like below:
- and then, I start down server,when the node recovery,the node get opertion can’t get value,the c sdk memory always increase,too.the server status like below:
Depending on how many operations you are scheduling, the C SDK’s memory usage may increase as it attempts to reschedule packets because of failed nodes. Can you perform an analysis or show some code indicating how you are scheduling operations? you should naturally expect some sort of temporary degradation of performance if a node is down, but this shouldn’t cause a leak per-se.
for example,I use “top” to monitor process.when all server up,the memory usage is 0.1%,and then one node down,memory increase to 3%,and then,the down node recovery,memory usage always keep 3% ,and end,one node down,the process not use already allocate memory, memory usage 3% to 3.1% very quickly.