Perhaps someone could help with the initial configuration of the network so I could get going. I’m probably making an elementary mistake. I simply want Red --> WAN, Green, offering DHCP addresses in 192.168.10.X range.
I am able to install Nethserver with a single interface connected to my present router (192.168.10.1), which gives DHCP addresses in the 192.168.10.2-255 range.
On bootup, I connect to the server via the web interface at the address the router has assigned (192.168.10.3) and set the “red” interface.
Then I set the green interface to “static” at 192.168.10.3. (But what should I set the gateway to be??)
Finally, I set the green interface to serve addresses via DHCP on the 192.168.10.1-255 range.
Now I unplug the router, plug the red interface directly to the internet, and plug my laptop into the green interface.
While I can access the web interface at 192.168.10.3, I can’t ping the network from the Nethserver console.
@mrmarkuz - thanks for your input. Since the WAN goes directly to the ISP, I thought better to try to get it working via my present router:
router --> red (acquire ip via DHCP); green (offer IP via DHCP) <–> laptop
While that works, speedtest via the console or web interface is 700Mb, while via my laptop it’s 25Mb. Would this be expected? The only thing I’ve added to the default install is fail2ban.
Hardware: Dell OptiPlex 980
CPU: 4 x Intel® Core™ i5 CPU 750 @ 2.67GHz
8GB RAM.
I’m not averse to upgrading my hardware, if that would make a difference.
@mrmarkuz - thanks for your input. After discovering the the linux driver for my thinkpad NIC’s are not giving me gigabit speeds, I switched to Windows where they are. So I’ve pared the issues down to one. When I do a speed test, I get this repeatedly on the Nethserver console:
e1000e 0000:00:19.0 em1: Detected Hardware Unit Hang:
TDH <24>
TDT <8a>
next_to_use <8a>
next_to_clean <24>
buffer_info[next_to_clean]:
time_stamp <100dfab1d>
next_to_watch <25>
jiffies <100dfcfb8>
next_to_watch.status <0>
MAC Status <40000083>
PHY Status <796d>
PHY 1000BASE-T Status <7800>
PHY Extended Status <2000>
PCI Status <10>
That em1 is the Dell adapter (Green). It’s supposed to be gigabit, but somehow I’m guessing that the driver is crashing. (My Red adapter is a PCI card with two interfaces.) Any thoughts?
For anyone else coming this way, there are clearly serious issues with the e1000e driver for Intel NIC’s. I will add more if I find a solution, but I am not sanguine. So far, bug reports have been ignored (see that link).
Fix broken e1000 driver
-----------------------
db networks setprop enp0s25 ethtool_opts "\"-K \${DEVICE} gso off gro off tso off\""
signal-event interface-update
Which seems to work for me.
Interestingly, my motherboard has 2 Intel NICs built in, an 82579LM and 82574L that both use the e1000e driver. It was only the 82579LM that had issues.
Thanks both. This is still very much in progress. I have already tried EddieA’s solution (see the link in my previous post). That didn’t help. What did help is turning off ASPM in BIOS. On my Dell running the Nethserver, that got me to ~ 0.7 Gb/s down and ~0.35 Gb/s up. (Though a windows machine on this network gets ~.95 down and ~1.05 up.)
On my laptops, I have similar Intel ethernet hardware, using the same versions of the e1000 drivers.
(Hardware is 82579LM rather than 82578DM on the Dell). There the download speeds are the same 0.7Gb/s, but on Thinkpads T520 and W520 with kernel 4.15.0, the upload speeds are precisely 4.16 Mb/s (!), and none of the reported tricks work. On a T530 with kernel 4.4.0, I get 0.7 down and 0.35 up, as with the Dell.