antminer-monitor icon indicating copy to clipboard operation
antminer-monitor copied to clipboard

Benchmark loading time

Open anselal opened this issue 6 years ago • 14 comments

I need feedback on how much time it takes to load the page with as many miners as possible, from a couple hundred to a few thousand.

@shotaatosh @danter83 ? Can you give me some feedback ?

You can post here the number of miners and the time it takes to load the page. Also if you have changed the socket.timeout this info would also be useful.

Any feedback would help !!!

PS: The loading time also depends on the network setup.

UPDATED: Also inactive miners add a significant delay to the page depending on the socket timeout parameter

anselal avatar Mar 28 '18 12:03 anselal

I can gather no more than 60 miners in one place for now

shotaatosh avatar Mar 28 '18 12:03 shotaatosh

@shotaatosh Fair enough

anselal avatar Mar 28 '18 12:03 anselal

updated to latest version, moved the db and everything is working, for now 44xS9s and 2xA3 also 44xD3s in another location, but I can't merge them fro now, but with VPN. loading time for S9s is quite fast, what exact info can I get for you?

shotaatosh avatar Mar 28 '18 13:03 shotaatosh

https://github.com/inconshreveable

loading time with a hugely mixed active-inactive list of 85 pcs of D3-s inactive: 49 active: 36 TTFB: 43.33sec

gnanet avatar Mar 29 '18 00:03 gnanet

I cant say how much time will it take as I need the version which supports port numbers so I can test - I have only remote access to my miners without access to LAN.

danter83 avatar Mar 29 '18 00:03 danter83

this ttfb was measured over a distance with a sshtunnel from home to the mining farm, with antminer-monitor running at the farm

gnanet avatar Mar 29 '18 00:03 gnanet

@gnanet If you remove the inactive miners the loading time with decrease significantly

anselal avatar Mar 29 '18 05:03 anselal

@anselal i know, but this was at least a real-world scenario, where last time i managed the ip list there were much more miners active, than dead or removed ;)

Is there a discussion about how to keep the list of miners uptodate, but not risking to miss alerts of failing miners. Kind of first alert about, then the failure could be acknowledged, at that point active monitoring shall not check the device, and after a configurable grace time the miner could be removed automatically.

gnanet avatar Mar 29 '18 16:03 gnanet

250 miners 1 inactive TTFB: 3.652 sec on server.

Pavelivn avatar Mar 30 '18 02:03 Pavelivn

@gnanet Notifications will be part of this software in the next releases.

@Pavelivn Nice !!! What do you think about the time ? I think it is pretty quick but can be quicker if I change my code !!!

anselal avatar Mar 30 '18 05:03 anselal

You can change the code provided that there are no false positives. Sometimes 1-2 miners are displayed inactive, although they are active.

Pavelivn avatar Mar 30 '18 05:03 Pavelivn

@Pavelivn I hear that for the first time. If the connection to the miner is lost then the miner is displayed as inactive. I guess if you refresh the page the miner appears as active again ?

anselal avatar Mar 30 '18 05:03 anselal

All right. There is a suggestion that miners, under certain conditions, respond with a delay.

Pavelivn avatar Mar 30 '18 05:03 Pavelivn

@Pavelivn If this happens to your setup often you can try to increase the socket timeout in the lib/pycgminer.py file

anselal avatar Mar 30 '18 05:03 anselal