| 08:07 | topi`> | can somebody help me configure the lan ports of my clearfogpro? |
| 08:07 | topi`> | I can make exactly one (1) port work from the switch, by assigning it an ip and doing ip link up dev eth1 ; ip link up dev lan1 |
| 08:08 | topi`> | but the other ports won't work |
| 08:08 | vpeter> | Didn't you already figure out the process? |
| 08:11 | topi`> | I figured it out; but it only enabled 1 out of 6 ports |
| 08:16 | vpeter> | I see. In openwrt all 6 ports are working :) |
| 08:16 | topi`> | does openwrt use a bridge device (br0)? |
| 08:17 | vpeter> | Yes if I remember correctly. But I can check at home |
| 08:53 | Ke> | yeah I saw some funny bridge hack in mcbin as well |
| 08:56 | topi`> | but mcbin doesn't have a built-in switch |
| 08:56 | topi`> | it's the switch that causes me problems |
| 11:13 | jnettlet[m]> | vpeter: fyi the 6th port can only run at 1gbps speed, it can't autonegotiate, and it is tied to the status of Port 4. In general we only consider ports 1-5 usable. |
| 11:13 | jnettlet[m]> | topi`: which kernel are you using? |
| 11:16 | vpeter> | jnettlet[m]: Yes, it is "special" port :) But it serves fine to me with carefully considering what is on the other side. |
| 11:17 | jnettlet[m]> | vpeter: just making sure you were aware. It is functional as long as you understand the limitations |
| 11:18 | vpeter> | :) |
| 14:57 | topi`> | jnettlet[m]: I think it's 4.14 kernel from igor pecovnik |
| 14:57 | topi`> | maybe it's lacking some special patches to make that Marvell switch chip work? but strangely lan1 wors? |
| 14:57 | topi`> | works |
| 14:58 | topi`> | I'm looking at some Turris Omnia guides, it has very similar hardware to clearfogpro |
| 15:00 | topi`> | the Omnia seems to be using Armada 385 |
| 15:14 | topi`> | is this dsa.txt relevant to Clearfog Pro? https://www.kernel.org/doc/Documentation/networking/dsa/dsa.txt |
| 15:16 | vpeter> | topi`: I think so. Because I was thinking to use this for my fried wan port. But then I got myself sfp module. |
| 15:20 | topi`> | hmm, a 10 Gbit SFP module would be nice |
| 15:23 | vpeter> | Well, my upload is only 70mbit :( |
| 15:28 | jnettlet[m]> | topi`: yes the switch is controlled via DSA |
| 15:29 | jnettlet[m]> | 10Gbit is nice. The switching equipment is just expensive. |
| 15:30 | jnettlet[m]> | Basically I have 1 MCbin that does 10G to my x86 build server, and 10G to my mcbin builder, and then 2.5Gb to a CFPro for the rest of the test network |
| 15:34 | jnettlet[m]> | vpeter: 70Mbps upload is not bad at all. |
| 15:35 | jnettlet[m]> | do you have a test case that I can use to try and reproduce the packet loss? |
| 15:35 | jnettlet[m]> | have you tested the interfaces with iperf? |
| 15:39 | vpeter> | jnettlet[m]: Sorry, this is DOWNLOAD. Upload is 2 mbit :( :( |
| 15:40 | jnettlet[m]> | yeah 2Mb is pretty rough. |
| 15:41 | vpeter> | jnettlet[m]: no test case. It happens random after powering cf pro. But not always. Sometimes rebooting helps, sometimes not. In this case I reboot cable modem too. And sometimes even this doesn't help. It only helps to wait 5 or 10 minutes and then everything is coming up just fine :-) |
| 15:42 | jnettlet[m]> | sounds like you are getting some sort ip address or mac address conflict. |
| 15:43 | jnettlet[m]> | possibly 2 dhcp servers on the same network? |
| 15:45 | vpeter> | Nothing like that. Maybe I forgot to wrote that only wan connection between cf and modem is problematic. |
| 15:47 | jnettlet[m]> | and this is using the SFP gigabit module? |
| 15:49 | vpeter> | Exactly. My next test is with DSA to rule this out. |
| 15:50 | jnettlet[m]> | next time it happens can you grab dmesg | grep -i sfp and post it. I am curious if it is a problem with the hotplugging |
| 17:52 | vpeter> | jnettlet[m]: https://pastebin.com/JBvqf4UX but now I don't have any problem. |
| 17:53 | jnettlet[m]> | that all looks correct. If the problem happens we can compare |
| 17:54 | vpeter> | I noticed this once but not sure if related to this problem. |
| 17:54 | vpeter> | mvneta f1034000.ethernet eth2: bad rx status 0f816000 (crc error), size=1536 |
| 19:27 | topi`> | [ 45.280053] mv88e6085 f1072004.mdio-mii:04 lan3: Link is Up - 1Gbps/Full - flow control rx/tx |
| 19:27 | topi`> | this looks like this kernel does have the necessary support for the marvell switch chip, right? |
| 19:34 | topi`> | can anyone provide me a script that would setup the cfp router ports I could test? |
| 19:35 | topi`> | or maybe I just go the easy way and plug a d-link gigabit switch to the lan1 port and be done with it |
| 19:35 | topi`> | but it kind of kills the meaning of the cf pro :) |
| 19:50 | topi`> | stupid question, but do I need to enable IPv4 forwarding in /proc/sys/net/ip_forward ?? |
| 20:26 | topi`> | this is really getting on my nerves, children crying and I stil cannot access the bloody NAS via cfpro |
| 20:26 | topi`> | RTNETLINK answers: File exists |
| 20:26 | topi`> | ifup: failed to bring up br0 |
| 20:26 | topi`> | this is the furthest I have gotten |
| 20:28 | topi`> | root@clearfogpro:/home/topi# brctl addif br0 lan2 |
| 20:28 | topi`> | device lan2 is already a member of a bridge; can't enslave it to bridge br0. |
| 20:28 | topi`> | oh, it is already there. Hmm, let's have lan4 then |
| 20:29 | topi`> | now lan2 and lan4 are bound to br0. If it works the way I think, I should now be able to ping machine A (connected to lan2 port) from machine B (connected to lan4) |
| 20:30 | jnettlet[m]> | topi`: really all our efforts have gone into supporting lede /openwrt. |
| 20:30 | topi`> | yeah but they're still linux and will use brctl and/or iproute2 underneath |
| 20:31 | jnettlet[m]> | If you install that everything will just work. Pretty early on we determined most Linux distributions were absolutely terrible at advanced networking. |
| 20:31 | topi`> | yeah that's true |
| 20:31 | topi`> | but I'm not going to give up debian for openwrt |
| 20:31 | topi`> | not until hell freezes |
| 20:31 | jnettlet[m]> | But you are using a random developers kernel. |
| 20:32 | topi`> | a good point |
| 20:32 | jnettlet[m]> | The lede kernel and patches are tested by us and going into consumer products. |
| 20:33 | jnettlet[m]> | We have stretch images posted for the CF platform |
| 20:33 | topi`> | right |
| 20:35 | jnettlet[m]> | The thing is Debian is crap for routing |
| 20:35 | topi`> | I understand that |
| 20:35 | topi`> | but the thing I'm trying to achieve is fairly simple |
| 20:35 | topi`> | like, dead simple |
| 20:36 | jnettlet[m]> | Really you should look at VyOS. Based on Debian, advanced routing and fully supports or hardware |
| 20:37 | topi`> | lots of debian forks lately |
| 20:39 | jnettlet[m]> | Well Debian is working hard to make it near impossible to contribute |
| 20:39 | topi`> | strange... lan2 is working now |
| 20:40 | topi`> | evidently I forgot to add lan2 to dhcpd's listening list |
| 20:42 | topi`> | sorry for bugging you guys.. I really should concentrate on beefing up my competence |
| 20:54 | topi`> | I can finally hook up my NAS, on lan port 5 |
| 20:55 | topi`> | LAN6 I can reserve for some special use cases, if any |
| 20:55 | topi`> | I wonder if I can use LAN6 with a linux box where I disable autonegotiate, force gigE and full duplex |