I was trying to create ~150-million metrics (pushing host to metric, lots
of hosts and metrics).
I started getting heap exceptions in OpenTSDB. I'm only using it to create
metrics via the API currently, I'm not throwing any data at it.
I started getting OOM / Out Of Memory errors in OpenTSDB. Does the
OpenTSDB implementation require keeping all UIDs in memory at the same time
which would cause me to run out of heap? I would have assumed it was an LRU
cache or something like that.