Emtu

Part of the issue is the Bell Hub 2000 doesn't really have a bridge mode. What it can do it passthrough the PPPoE credentials so you can connect your own router to a LAN port on the Bell Hub, then have the Bell Hub connected to the Nokia ONT without running into double NAT or needing to configure VLAN 35. This isn't really an ideal setup if the Bell Hub is causing issues though - and on top of that you still need your PPPoE credentials anyway. Best bet is your second option and completely bypass the Bell router.

I have a similar setup at one of the businesses provide IT services for. One thing to note is if you're on Bell Aliant (Atlantic Canada) this doesn't apply. First there's two pieces of information you're missing which is why you just can't connect your own router to the Nokia ONT. 1. You need your PPPoE credentials. These are different from your Bell Hub login credentials. Assuming you're on a business account it would be best to call Bell and have them provide these. The user ID will be XXXXXXXX@bellnet.ca or possibly b1XXXXXX. 2. From the Nokia ONT Bell uses VLAN 35. From there you just need to configure your router WAN interface to use VLAN 35 as well as PPPoE with the credentials provided by Bell. You didn't mention what router you intend to use or what the internet speed is. You might need some tweaking if you're aiming to hit 1Gbit symmetrical for example. PPPoE performance can be hit or miss depending on the router brand/model.

Assuming you've double and triple checked that it isn't just a simple cabling issue I'd first try verifying the H240 can detect a directly attached disk. If that works then the backplane is most likely the issue. That backplane is old, check if it's running the latest available firmware from HP. It looks like it's a 3Gbit SAS expander as well. While it technically should just negotiate at the appropriate speed it might just be fickle about the host device - especially with a much newer 12Gbit card. Some of these 1.5/3.0Gbit SAS expanders weren't as forgiving with card compatibility like we're used to nowadays.

It's a shame you didn't show us even just a quick and dirty dd benchmark for kicks.

Believe it or not this drive is actually built with Canadian developed technology. That Novachips controller and the memory on the cards are the product of a company originally based in Ottawa which had their chip development division (called HLNAND and for a short time NDS Devices) bought by Novachips, a Korean company. I worked for the parent company before and after the purchase (not the chip division), but I actually still have a prototype SSD sporting the earlier NVS3600A controller. The idea behind the controller was the same as the LTT video demonstrates as it wasn't necessarily performance oriented but scalability oriented. In 2013 they were showing off 3.5" SSDs with 7.2TB of storage for example, and by 2015 they had a very long 32TB PCIe card absolutely packed with their HLNAND flash. One of the older 32TB SSDs from their website: And the prototype SSD which has the NV3600A controller but uses Intel 29F16B08CCME2 NAND:

For an extra £25 you could get a brand new 2TB Barracuda with a warranty and DOA protection from scan.co.uk. Is saving £25 really worth the potential headache of a used drive? Pretty sure the EARX WD Green drives are 5400RPM not 7200 as well.

Depending on what is actually being hosted a VPN may not even be necessary. I'm not really sure how you've come the the conclusion that a VPN is less hassle to setup and manage than just port forwarding.

Curious to know if anyone has actually got this to to work 100%, my cached downloads work great but non-cached downloads are still no good, usually topping around 10MB/s but eventually slowing down to under 1MB/s within minutes. I've tried Ubuntu 18.04, 16.04, Centos 7-1804, tried changing the storage from mdadm to btrfs, tried changing the docker network driver from bridge to macvlan. No improvement, on the cache-build download the cpu/disk/memory utilization is all low. Someone had similar issues - second post here: (https://github.com/steamcache/steamcache/issues/15) I don't doubt someone has got this working but with low hardware utilization across the board I suspect I'm running into a software or docker bug.

Not sure what's going on with mine, it's caching and working as it should - except download performance is total butt. I just can't figure out where it's being held up, I'm suspecting a CPU bottleneck but I'm not sure why. It's been assigned two cores of a dual L5639 system - increasing core count doesn't improve the performance. Download speed without cache server (maxes out internet connection): Download speed with cache server (before game has been cached, it is creating the cache): Disk speed available to the cache server: CPU usage of the cache server during download (creating cache) - they stay at low percentage: CPU usage as reported by ESX - the server is virtualized and on a 10G network: Really odd flatline happening which is indicative of a bottleneck. But on the OS of the caching server it reports low CPU usage. I'm unfamiliar with Docker so I wonder if there's some kind of tweak that needs to be applied but maybe someone else has some input?