Re: [LARTC] multipath device round robin not working?
On Sat, Jan 13, 2007 at 12:54:24PM +0100, [EMAIL PROTECTED] wrote: Hi, I have a linux server running kernel 2.6.19 that is connected with 2 seperate 100Mbit links to the same isp: +---+ +---+ | I | +---+ | | | S | | | |eth0 --+--+ P | | | | | | S | | | | linux 2.6.19 | | W || ISP GATEWAY | | | | I | | | |eth1 --+--+ T | | | | | | C | | | +---+ | H | +---+ +---+ Both links have their own ip but have the same gateway. The problem is I can't seem to get egress traffic load balanced over the 2 nics. IP config after boot (dhcp from isp) ip a: 1: lo: LOOPBACK,UP,1 mtu 16436 qdisc noqueue link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 inet 127.0.0.1/8 brd 127.255.255.255 scope host lo 2: eth0: BROADCAST,MULTICAST,NOTRAILERS,UP,1 mtu 1500 qdisc pfifo_fast qlen 1000 link/ether 00:00:00:00:00:0f brd ff:ff:ff:ff:ff:ff inet 10.0.0.110/24 brd 10.0.0.255 scope global eth0 3: eth1: BROADCAST,MULTICAST,NOTRAILERS,UP,1 mtu 1500 qdisc pfifo_fast qlen 1000 link/ether 00:00:00:00:00:ed brd ff:ff:ff:ff:ff:ff inet 10.0.0.120/24 brd 10.0.0.255 scope global eth1 Default routing table after boot ip r: 10.0.0.0/24 dev eth0 scope link 10.0.0.0/24 dev eth1 scope link metric 1 127.0.0.0/8 dev lo scope link default via 10.0.0.1 dev eth0 default via 10.0.0.1 dev eth1 metric 1 I enabled ip_forward and set arp_ignore to 1 for eth0 and eth1 to make sure the correct nic answers to arp requests. I tried to get the egress load balancing to work by replacing the above two default routes with: ip route add default mpath drr nexthop via 10.0.0.1 dev eth0 weight 1 onlink nexthop via 10.0.0.1 dev eth1 weight 1 onlink I assumed that with mpath device round robin both nics would be used more or less equally, but the reality is only one of the nics actually works and the second nic even stops responding to arp requests. Am I doing something totally wrong or impossible here or is the device round robin code not working properly? Curiosity but why use such a setup is your ISP link 2Gbp/s ? Why not bond if you want HA. why its not round robining. I am going to guess but this line default via 10.0.0.1 dev eth0 costs less to use than default via 10.0.0.1 dev eth1 metric 1 so it should never use the second. I say guess cause I don't know what the default metric is if you do add one. What you want it to look something like is default proto static metric 5 nexthop via 144.132.144.1 dev vlan2 weight 5 nexthop via 10.20.20.230 dev ppp0 weight 20 There is a link to a howto on the web site that steps out how to set this up Alex ___ LARTC mailing list LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc signature.asc Description: Digital signature ___ LARTC mailing list LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
Re: [LARTC] Linux as T1 router
Greetings, : I am thinking about using a linux server as a T1 router. I have : searched the list, but have not found a discussion about what I'm : trying to do. I have a situation where the Cisco router I'm using : will not handle the additional bandwidth I added recently. : Unfortunately, I cannot afford the Cisco unit that will. I would : like to know if anyone has successfully done this. I have been : looking at the Sangoma T1 cards. Would anyone be so kind as to : share their experience in this area. Any advice would be much : appreciated. I can recommend the Sangoma T1 cards. I have been using the S508 (ISA) and S514 (PCI) models since 1999. These cards and the (open source) drivers and management software are easy to use. The company is responsive and supportive of their product. The Sangoma crew have worked over the years to contribute their drivers into the stock kernel, so it is likely (unless the card you choose is a newly released card) that your card will be supported by your default distribution of choice. The software management tools are provided by a separate package, including tools for configuring the (optional) onboard CSU/DSU and diagnosing the frames received by the unit. Best of all, I can report that I have only ever found one bug in working with their software and drivers, and this was a corner-case bug that they had identified before I reported it to them (several years ago). In short, the software and hardware is very reliable. -Martin -- Martin A. Brown http://linux-ip.net/ ___ LARTC mailing list LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
[LARTC] Linux as a multicast router
Hi all, We want linux (ubuntu) with 2.6 kernel to act as multicast router and to pass multicast packets between different subnets. And linux machine as router has two network cards having two different subnets assigned onto. We first worked sysctl -w net.ipv4.ip_forward=1 line on linux shell that made linux to work as unicast router. (like host, the multicast packets were discarded at NIC). Then at weekend we heard about sysctl -w net.ipv4.conf.all.mc_forward=1 line that is said to satisfy multicast routing requirement, but we did not try this yet. Is this last line enough to work linux as multicast router ? There is also one alternative we found: XORP ( Open Source IP Router ) http://www.xorp.org/livecd.html#getting But this works from live cd. We want to also use linux shell at the same time. And also we did not try this tool yet. It is enough to use the linux as router, we do not consider the router algorithms or any other thing. Thanks in advance... Emin ___ LARTC mailing list LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
Re: [LARTC] dst cache overflow (bridged wan interfaces) [appears to be SOLVED]
Yes, the problem appears to be in the http://www.ssi.bg/~ja/#routes patch. Perhaps this patch is to any purposes and break the routes table or something about it. I tested for 2 days now with 4 pcs and amule/azureus configured very agressively to enable connections quickly and monitorized with rtstat util (as someone point to me) and I had seen how is working the routing in the linux box more numerically. I had take sense on all the comments about this problem and how to optimize the routing, ¡¡THANKS TO ALL!! Appears I can run fine with this configuration and use in production environment, but I'll wait for 2 or 3 weeks before pass it into production. Thanks to all!! El Vie, 12 de Enero de 2007, 2:13, ArcosCom Linux User escribió: The problem appears to be in the routes patch (after 1 day with 1 workstation with amule configured very agresively). I'm trying now the 2.6.19.2 kernel with the configuration exposed here, I'll tell you if the problem were (or not) the patch for dead-gw-detection/multipath-routes from nano-howto. Perhaps this patch is for specific configuration and need more accurate routes config (don't know). As I said: I'll say if I the problem persist in some days. Thank you very much. Regards El Mie, 10 de Enero de 2007, 21:14, ArcosCom Linux User escribió: I recompiled yet 2.6.19.1 kernel (using iptables with the same patches too). The configuration for this test is: 1) linux box with 2.6.19.1 kernel (SMP machine) with these patches/modules: a) l7-filter b) ipp2p c) connlimit d) set 2) 4 ethernet interfaces: a) 2 external (eth1 and eth3) interfaces with balanced links (as described in nato-howto) bridged as wan0 with static IPs assigned to wan0 and wan0:1 b) 2 internal ineterfaces (eth0 and eth2) in bridge zlan0 with STP enabled and configured. IPTABLES relevant configuration: # iptables -t nat -vn -L POSTROUTING Chain POSTROUTING (policy ACCEPT 185 packets, 16649 bytes) pkts bytes target prot opt in out source destination 26 1529 MASQUERADE 0-- * wan010.1.1.0/27 0.0.0.0/0 0 0 MASQUERADE 0-- * wan0:1 10.1.1.0/27 0.0.0.0/0 ROUTES CONFIGURATION: # service rt status === REGLAS DE ENRUTAMIENTO === 0: from all lookup local 50: from all lookup main 151:from NET_PUB1 lookup 151 152:from NET_PUB2 lookup 152 220:from all lookup 220 32766: from all lookup main 32767: from all lookup default === TABLAS DE RUTAS === === MAIN === NET_PUB1/26 dev wan0 proto kernel scope link src IP_PUB1 NET_PUB2/24 dev wan0 proto kernel scope link src IP_PUB2 192.168.3.0/24 dev zlan0 proto kernel scope link src 192.168.3.247 192.168.2.0/24 dev zlan0 proto kernel scope link src 192.168.2.247 192.168.1.0/24 dev zlan0 proto kernel scope link src 192.168.1.247 10.1.1.0/24 dev zlan0 proto kernel scope link src 10.1.1.6 169.254.0.0/16 dev zlan0 scope link 239.0.0.0/8 dev zlan0 scope link === wan0 TABLA 151 === default via GW_PUB1 dev wan0 proto static src IP_PUB1 prohibit default proto static metric 1 === wan0 TABLA 152 === default via GW_PUB2 dev wan0 proto static src IP_PUB2 prohibit default proto static metric 1 === TABLA 220 (defecto) === default proto static nexthop via GW_PUB1 dev wan0 weight 1 nexthop via GW_PUB2 dev wan0 weight 1 ROUTING parameters configuration: # grep . /proc/sys/net/ipv4/route/* /proc/sys/net/ipv4/route/error_burst:5000 /proc/sys/net/ipv4/route/error_cost:1000 grep: /proc/sys/net/ipv4/route/flush: Operación no permitida /proc/sys/net/ipv4/route/gc_elasticity:8 /proc/sys/net/ipv4/route/gc_interval:60 /proc/sys/net/ipv4/route/gc_min_interval:0 /proc/sys/net/ipv4/route/gc_min_interval_ms:500 /proc/sys/net/ipv4/route/gc_thresh:32768 /proc/sys/net/ipv4/route/gc_timeout:300 /proc/sys/net/ipv4/route/max_delay:10 /proc/sys/net/ipv4/route/max_size:524288 /proc/sys/net/ipv4/route/min_adv_mss:256 /proc/sys/net/ipv4/route/min_delay:2 /proc/sys/net/ipv4/route/min_pmtu:552 /proc/sys/net/ipv4/route/mtu_expires:600 /proc/sys/net/ipv4/route/redirect_load:20 /proc/sys/net/ipv4/route/redirect_number:9 /proc/sys/net/ipv4/route/redirect_silence:20480 /proc/sys/net/ipv4/route/secret_interval:600 When I test it along some weeks with intensive traffic I'll put here more info about this test. If somebody has any idea on how to solve the problem, please, tell us. I'm a bit desesperate with this issue. Regards ___ LARTC mailing list LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc ___ LARTC mailing list LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc
[LARTC] Re: Example on using fwmark with masks. Please help!!
Any help? El Vie, 12 de Enero de 2007, 21:26, ArcosCom Linux User escribió: Hi, could anyone provides any example about the use of ip route command to force the use of one route using masks in the mark? The configuration is: 1 LAN (zlan0) iface N WAN (wan0 ... wanN) ifaces with static IPs and load balanced. iptables 1.3.7 kernel 2.6.19.2 iproute 2.6.19 I'm yet setting marks into packets for QoS and its working, I now want to set some bits (OR) at the end of the mark. For example, I want to use 0x8000 to add another mark to the packet for routing. The packet is market yet with a QoS mark (--set-mark), 0x5 (for example). I need: 1) The packet been marked with 0x8000 OR 0x0005 = 0x8005 2) Route the packet with 0x8005 AND 0x8000 = 0x8000 over wan0 (for example) 3) Classify the packet with 0x8005 AND 0x0005 into wan0 1:4 class (for example) I know how to do this not having sense about MASKs, but ... Could anybody put here how to do it with them? Another question: What is the length of the mask? 16bit? 32bit? Thanks!! ___ LARTC mailing list LARTC@mailman.ds9a.nl http://mailman.ds9a.nl/cgi-bin/mailman/listinfo/lartc