Re: [Server-devel] F9 port
Martin Langhoff wrote: On Mon, Aug 18, 2008 at 4:49 AM, Jerry Vonau [EMAIL PROTECTED] wrote: Well, I received the AA's. As a quick test I plugged them into my F9 box, after using the olpc ifcfg-files as a template, with very little fiddling I think I have them running under F9. I'll post the diff later. More testing when the xo gets here, or my friend with a laptop. Excellent. If you have a normal laptop, you'll need the recent 802.11s kernel modules. What does need work is the network_config file, and the idea of running it on boot, you need to pass the server_number to the script right? Can't do that on boot, and by default on F9 and the livecd there is no ifcfg-eth0 file to work with anyway. Here are the challenges: - network_config runs on firstboot, and can be re-run later if network interfaces are added/removed Think if the udev rule use different names from the default names, then the devices won't flip around. ie msh0=mesh0 eth1=lan using something like: KERNEL==eth*, ATTR{address}==00:52:8b:d5:04:48, NAME=lan Then the ifcfg-files could use the new alias names. Thoughts? - network_config sets up - eth0 for the WAN - ethN where N0 and the interface is wired as LAN ports - the ethX/mshX pairs as LAN+Mesh ports - and appropriate bridging across the LAN ports, routing/NAT'ting between LAN and WAN This there a reason for the multi-bridges? Could you not just add the msh devices to a single bridge and adjust the subnetmask to fit? Just wondering... On a side note can I get the output of brctl show, ip route, iwconfig from a running installation, to compare with please. The debugging web page is a good start, but lacking the bridging info, you really need to see both the routing and bridging to get a better picture. - udev scripts so that once an ethX/mshX is set, things work well even if the machine boots w/o the AA, or if the AA is plugged _after_ we've booted 75-persistent-net-generator.rules takes care of the naming, what you want is an auto ifup dev with the plugging in of the usb dev? - possibly other bits that I'm forgetting :-) I heard rumbling about xo name resolution, how about dhcpd auto-updating named? If this xs-conf is meant to be a rpm then would the *_config files not live better in /sbin and the ifcfg-* and *.conf file templates in /usr/share/xs-config? Yes. And all that symlink mess get deprecated. I'm working on that too :-) That will take a bit of time... ;-) Jerry ___ Server-devel mailing list Server-devel@lists.laptop.org http://lists.laptop.org/listinfo/server-devel
Re: [Server-devel] F9 port
Jerry Vonau wrote: Martin Langhoff wrote: On Mon, Aug 18, 2008 at 4:49 AM, Jerry Vonau [EMAIL PROTECTED] wrote: Well, I received the AA's. As a quick test I plugged them into my F9 box, after using the olpc ifcfg-files as a template, with very little fiddling I think I have them running under F9. I'll post the diff later. More testing when the xo gets here, or my friend with a laptop. Excellent. If you have a normal laptop, you'll need the recent 802.11s kernel modules. What does need work is the network_config file, and the idea of running it on boot, you need to pass the server_number to the script right? Can't do that on boot, and by default on F9 and the livecd there is no ifcfg-eth0 file to work with anyway. Here are the challenges: - network_config runs on firstboot, and can be re-run later if network interfaces are added/removed Think if the udev rule use different names from the default names, then the devices won't flip around. ie msh0=mesh0 eth1=lan using something like: KERNEL==eth*, ATTR{address}==00:52:8b:d5:04:48, NAME=lan Then the ifcfg-files could use the new alias names. Thoughts? - network_config sets up - eth0 for the WAN - ethN where N0 and the interface is wired as LAN ports - the ethX/mshX pairs as LAN+Mesh ports - and appropriate bridging across the LAN ports, routing/NAT'ting between LAN and WAN Well, that is interesting if I boot with rhgb enabled, network start fails horribly... Disabling rhgb, things work perfect. :-) Think the above idea of installing an alias of the real devices to use via udev, for the ifcfg files should work. If I can set the alias of the devices that are in use, then the config files for the network layout could be fixed. The names of the ifcfg files become static, mesh1 mesh2 mesh3,mshX ethX wless1, wless2, wless3, wired1 wan1, wan2(?), bridges would remain as is. Feedback? This there a reason for the multi-bridges? Could you not just add the msh devices to a single bridge and adjust the subnetmask to fit? Just wondering... On a side note can I get the output of brctl show, ip route, iwconfig from a running installation, to compare with please. The debugging web page is a good start, but lacking the bridging info, you really need to see both the routing and bridging to get a better picture. - udev scripts so that once an ethX/mshX is set, things work well even if the machine boots w/o the AA, or if the AA is plugged _after_ we've booted 75-persistent-net-generator.rules takes care of the naming, what you want is an auto ifup dev with the plugging in of the usb dev? On my test box, installing a second wired nic was automatically given eth3 after configuring the mesh setup previously. Think all I need to do is parse the 70-persistent-net.rules for msh, eth matching on the mac address to use for the bases of the alias name. Jerry ___ Server-devel mailing list Server-devel@lists.laptop.org http://lists.laptop.org/listinfo/server-devel
Re: [Server-devel] F9 port
On Mon, Aug 18, 2008 at 11:26 PM, Jerry Vonau [EMAIL PROTECTED] wrote: Here are the challenges: - network_config runs on firstboot, and can be re-run later if network interfaces are added/removed Think if the udev rule use different names from the default names, then the devices won't flip around. ie msh0=mesh0 eth1=lan using something like: KERNEL==eth*, ATTR{address}==00:52:8b:d5:04:48, NAME=lan Then the ifcfg-files could use the new alias names. Thoughts? That would mean creating udev rules when network_config is run? That would be fine with me. Will it lead to having the eth and msh devices for the USB antennas lined up so that eth4 goes with msh4? - network_config sets up - eth0 for the WAN - ethN where N0 and the interface is wired as LAN ports - the ethX/mshX pairs as LAN+Mesh ports - and appropriate bridging across the LAN ports, routing/NAT'ting between LAN and WAN This there a reason for the multi-bridges? Could you not just add the msh devices to a single bridge and adjust the subnetmask to fit? Just wondering... I'm not sure. Wad might have an opinion... - udev scripts so that once an ethX/mshX is set, things work well even if the machine boots w/o the AA, or if the AA is plugged _after_ we've booted 75-persistent-net-generator.rules takes care of the naming, what you want is an auto ifup dev with the plugging in of the usb dev? Yes! I heard rumbling about xo name resolution, how about dhcpd auto-updating named? That's desirable, but I'm not sure how long we'll want to stick to BIND. I'm considering dnsmasq or djbdns. If this xs-conf is meant to be a rpm then would the *_config files not live better in /sbin and the ifcfg-* and *.conf file templates in /usr/share/xs-config? Yes. And all that symlink mess get deprecated. I'm working on that too :-) That will take a bit of time... ;-) getting there...! m -- [EMAIL PROTECTED] [EMAIL PROTECTED] -- School Server Architect - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Server-devel mailing list Server-devel@lists.laptop.org http://lists.laptop.org/listinfo/server-devel
Re: [Server-devel] F9 port
On Tue, Aug 19, 2008 at 6:04 AM, Jerry Vonau [EMAIL PROTECTED] wrote: Jerry Vonau wrote: Well, that is interesting if I boot with rhgb enabled, network start fails horribly... Disabling rhgb, things work perfect. :-) Odd. We definitely don't have a use for rhgb... Think the above idea of installing an alias of the real devices to use via udev, for the ifcfg files should work. If I can set the alias of the devices that are in use, then the config files for the network layout could be fixed. The names of the ifcfg files become static, mesh1 mesh2 mesh3,mshX ethX wless1, wless2, wless3, wired1 wan1, wan2(?), bridges would remain as is. Feedback? Less moving parts == more resilient -- I'm 100% for that :-) On my test box, installing a second wired nic was automatically given eth3 after configuring the mesh setup previously. Think all I need to do is parse the 70-persistent-net.rules for msh, eth matching on the mac address to use for the bases of the alias name. Sounds good - I'll let you do your magic. m -- [EMAIL PROTECTED] [EMAIL PROTECTED] -- School Server Architect - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Server-devel mailing list Server-devel@lists.laptop.org http://lists.laptop.org/listinfo/server-devel
Re: [Server-devel] F9 port
On Mon, Aug 18, 2008 at 11:26 PM, Jerry Vonau [EMAIL PROTECTED] wrote: On a side note can I get the output of brctl show, ip route, iwconfig from a running installation, to compare with please. ## XS with 2 nics, 1 AP, no AA. # brctl show bridge name bridge id STP enabled interfaces br0 8000. no br1 8000. no br2 8000. no # ip route 192.168.2.0/24 dev eth0 proto kernel scope link src 192.168.2.178 172.18.0.0/23 dev eth1 proto kernel scope link src 172.18.0.1 172.18.14.0/23 dev br2 proto kernel scope link src 172.18.14.1 172.18.12.0/23 dev br1 proto kernel scope link src 172.18.12.1 172.18.10.0/23 dev br0 proto kernel scope link src 172.18.10.1 172.18.64.0/21 via 172.18.1.8 dev eth1 172.18.8.0/21 via 172.18.1.1 dev eth1 scope link 172.18.16.0/21 via 172.18.1.2 dev eth1 172.18.24.0/21 via 172.18.1.3 dev eth1 172.18.32.0/21 via 172.18.1.4 dev eth1 172.18.40.0/21 via 172.18.1.5 dev eth1 172.18.48.0/21 via 172.18.1.6 dev eth1 172.18.56.0/21 via 172.18.1.7 dev eth1 169.254.0.0/16 dev br2 scope link default via 192.168.2.1 dev eth0 # iwconfig lono wireless extensions. eth1 no wireless extensions. eth0 no wireless extensions. br0 no wireless extensions. br1 no wireless extensions. br2 no wireless extensions. ### XS with 1 NIC, 1 AA # brctl show bridge name bridge id STP enabled interfaces br0 8000.005043280ad4 no msh0 br1 8000. no br2 8000. no 192.168.2.0/24 dev eth0 proto kernel scope link src 192.168.2.102 172.18.14.0/23 dev br2 proto kernel scope link src 172.18.14.1 172.18.12.0/23 dev br1 proto kernel scope link src 172.18.12.1 172.18.10.0/23 dev br0 proto kernel scope link src 172.18.10.1 172.18.64.0/21 via 172.18.1.8 dev dummy0 172.18.8.0/21 via 172.18.1.1 dev dummy0 172.18.16.0/21 via 172.18.1.2 dev dummy0 172.18.24.0/21 via 172.18.1.3 dev dummy0 172.18.32.0/21 via 172.18.1.4 dev dummy0 172.18.40.0/21 via 172.18.1.5 dev dummy0 172.18.48.0/21 via 172.18.1.6 dev dummy0 172.18.56.0/21 via 172.18.1.7 dev dummy0 172.18.0.0/20 dev dummy0 proto kernel scope link src 172.18.0.1 169.254.0.0/16 dev br2 scope link default via 192.168.2.1 dev eth0 # iwconfig lono wireless extensions. eth0 no wireless extensions. eth1 IEEE 802.11b/g ESSID:school-mesh-0 Nickname:schoolserver Mode:Ad-Hoc Frequency:2.412 GHz Cell: 02:25:75:14:3D:0E Bit Rate:54 Mb/s Tx-Power=13 dBm Retry limit:8 RTS thr=2347 B Fragment thr=2346 B Encryption key:off Power Management:off Link Quality=99/100 Signal level=-256 dBm Noise level=-96 dBm Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:1004 Tx excessive retries:5 Invalid misc:54 Missed beacon:0 msh0 IEEE 802.11b/g ESSID:mesh Nickname:Mesh Mode:Repeater Frequency:2.412 GHz Bit Rate:54 Mb/s Tx-Power=13 dBm Retry limit:8 RTS thr=2347 B Fragment thr=2346 B Encryption key:off Power Management:off Link Quality=39/100 Signal level=-81 dBm Noise level=-94 dBm Rx invalid nwid:0 Rx invalid crypt:0 Rx invalid frag:3290696 Tx excessive retries:30836 Invalid misc:339025 Missed beacon:0 dummy0no wireless extensions. br0 no wireless extensions. br1 no wireless extensions. br2 no wireless extensions. -- [EMAIL PROTECTED] [EMAIL PROTECTED] -- School Server Architect - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Server-devel mailing list Server-devel@lists.laptop.org http://lists.laptop.org/listinfo/server-devel
[Server-devel] F9 port
Well, I received the AA's. As a quick test I plugged them into my F9 box, after using the olpc ifcfg-files as a template, with very little fiddling I think I have them running under F9. I'll post the diff later. More testing when the xo gets here, or my friend with a laptop. What does need work is the network_config file, and the idea of running it on boot, you need to pass the server_number to the script right? Can't do that on boot, and by default on F9 and the livecd there is no ifcfg-eth0 file to work with anyway. If this xs-conf is meant to be a rpm then would the *_config files not live better in /sbin and the ifcfg-* and *.conf file templates in /usr/share/xs-config? On a second machine I used the current livecd to install (that could use a bit of work), once rebooted, I pointed the yum repos to F9 and my custom repo and ran yum upgrade. The custom repo has just a rpm to fake-out the deps the are in the current install. http://members.shaw.ca/jvonau/pub/xs-fixer-1.0.0-5.src.rpm Things went well, and upon reboot, I have a running olpc F9 box, I just have to migrate the ifcfg- config files from my workstation to the server, and continue testing. Looks good so far. More later, Jerry ___ Server-devel mailing list Server-devel@lists.laptop.org http://lists.laptop.org/listinfo/server-devel
Re: [Server-devel] F9 port
On Mon, Aug 18, 2008 at 4:49 AM, Jerry Vonau [EMAIL PROTECTED] wrote: Well, I received the AA's. As a quick test I plugged them into my F9 box, after using the olpc ifcfg-files as a template, with very little fiddling I think I have them running under F9. I'll post the diff later. More testing when the xo gets here, or my friend with a laptop. Excellent. If you have a normal laptop, you'll need the recent 802.11s kernel modules. What does need work is the network_config file, and the idea of running it on boot, you need to pass the server_number to the script right? Can't do that on boot, and by default on F9 and the livecd there is no ifcfg-eth0 file to work with anyway. Here are the challenges: - network_config runs on firstboot, and can be re-run later if network interfaces are added/removed - network_config sets up - eth0 for the WAN - ethN where N0 and the interface is wired as LAN ports - the ethX/mshX pairs as LAN+Mesh ports - and appropriate bridging across the LAN ports, routing/NAT'ting between LAN and WAN - udev scripts so that once an ethX/mshX is set, things work well even if the machine boots w/o the AA, or if the AA is plugged _after_ we've booted - possibly other bits that I'm forgetting :-) If this xs-conf is meant to be a rpm then would the *_config files not live better in /sbin and the ifcfg-* and *.conf file templates in /usr/share/xs-config? Yes. And all that symlink mess get deprecated. I'm working on that too :-) cheers, m -- [EMAIL PROTECTED] [EMAIL PROTECTED] -- School Server Architect - ask interesting questions - don't get distracted with shiny stuff - working code first - http://wiki.laptop.org/go/User:Martinlanghoff ___ Server-devel mailing list Server-devel@lists.laptop.org http://lists.laptop.org/listinfo/server-devel