Topic: Networking issue (Read 54 times)

Thanks for advice i shall take a look at the Ubiquiti. Do switch quantity and  sizes matter behind the router ?

I can also recommend the Ubiquiti EdgeRouter as it is a pretty powerful commercial grade router at a SOHO price level. Much better than anything you will get from Linksys, Netgear and the like, especially with that many rigs.

I run a number of rigs myself and also think just statically assigning them an IP address is the best option. My internal network IP scheme I use a /23 block (512 addresses) and keep the lower half of the block for my non-mining devices, and the upper half for mining devices. This way I could name my miner according to their IP address. So my rig named miner01 would get the address 192.168.1.1, miner02 would use 192.168.1.2, and so on. So between setting them up as static addresses not needing DHCP and having a simple naming and IP addressing scheme, this works out well for managing larger (50+) rig type environments.

By segmenting your network this way and with a capable router such as the Ubiquiti, you can even go a step further and apply strict firewall rules on your mining segment to further protect your network. Since you miners only need to connect to a handful of Internet based servers, a handful of firewall rules will help keep any unwanted traffic from going in/out to your rigs. I would recommend applying some rules to your normal LAN traffic as well, but that is getting a bit outside the topic.

How have the past routers died? Your network topology should be ok.

By my calculations, you’re currently running 25 switches at least? With 150 machines you should still be well within a single LAN CIDR block (192.168.0.1/24) so it’s not that you’re running out of addresses (you should be using ~180 addresses out of a possible 255 assuming a total of 25 switches and 155 mining rigs, this does not include other devices on the network like phones, personal computers, TVs, etc)


One possible issue is that if machines were going down enough times within the DHCP lease period (let’s say it’s 6 hours) your old leases might not be releasing causing new machines to not be able to establish a new lease. Assuming you have a 40 Address buffer in your CIDR block and you have a 6 hour DHCP lease, you could theoretically run out of LAN IPs if you have a machine turnover >40 in that timeframe (would need to be a reboot to lose the lease).

There are a couple ways around this.

The easier and most cost effective way would be to give all devices on your network a static IP (assuming you don’t) to solve the turnover assuming that’s the issue.

Another way would be to get a router that easily supports multiple downstream LANs and run more than one. Personally I run a Ubiquiti EdgeRouter with 2 downstream LANs (192.168.1.1/24 and 192.168.2.1/24) to keep my mining rigs on a separate LAN from my personal device WiFi (where having DHCP is important).

microtik

Hi,
i have come across a networking issue due to too many computers on the same router+switches. Two routers have died so far in the span of two months I have 50 rigs at the moment, and will expand to 150+. The issue right now is that the router does not asign IP's to new rigs. I have a linksys ea6350, dont know if it's too weak or switches are messing things up.

The setup looks like this : modem - > router -> 16port switch - > 8port switches

Any advice on wich router to buy, that would be able to handle this many computer? (without goin crazy with 2k eur + routers).

Thanks