[proxy-web] Hashrate graph craters on restart, very slow to recover

[vnprc]

This one seems very tricky. I've been banging my head on the desk for a week trying to fix it.

Every time I restart the server, the reported hashrate craters to zero and incrementally recovers over the next 15-30 minutes.

The flatline datapoints in this graph are:

time	hashrate
11:45	1.00 KH/s
11:46	1.42 KH/s
11:47	2.09 KH/s
11:48	9.85 KH/s
11:49	30.40 KH/s
11:50	78.28 KH/s
11:51	19.03 KH/s
11:52	24.67 KH/s
11:53	126.97 KH/s
11:54	418.03 KH/s
11:55	2.13 MH/s
11:56	6.22 MH/s
11:57	75.69 MH/s
11:58	285.54 MH/s
11:59	1.97 GH/s
12:00	9.83 GH/s

[vnprc]

Here is my most recent attempt to fix it. Still didn't seem to help. I give up. ¯\_(ツ)_/¯ I need to focus on more important things. https://github.com/vnprc/hashpool/blob/master/docs/WARM-START-TIME-SERIES-INITIALIZATION.md

[vnprc]

It seems to slowly climb for a while and then if a miner reconnects it gets knocked back down and starts the climb over again. This may be a critical clue. I think maybe it's doing the calculations based on a single miner connection and somehow a reconnection resets this process.

It should be graphing the aggregate hashrate of all miners and not only a single connection.