Closed mkovatsc closed 8 years ago
The MID problem is caused when they are reused before their lifetime ended (i.e., because the endpoints send with a rate higher than the maximum of 2^16/247s).
Looks like a false alarm, since the MID problem is caused by something else and the memory usage is simply enormous at high throughput... closing.
When logging the latency percentiles of servers with very high throughput, CoAPBench starts to mix up MIDs and randomly causes OutOfMemory exceptions.
Current workaround is to use "-Xms8192m -Xmx8192", however, sometimes failures still occur and not all machines have >>8 GiB RAM...