This is a discussion on Help : Load Balancing trouble within the Linux Networking forums, part of the Linux Forums category; Hello, I have a load balanced linux router working (thanks to lartc website). In fact, It's working too much ...
|
|||||||
| FAQ | Members List | Calendar | Search | Today's Posts | Mark Forums Read |
|
|||
|
Hello,
I have a load balanced linux router working (thanks to lartc website). In fact, It's working too much well ..... Some protocols, like IM protocols (ICQ, MSN) need to keep the same route with the server ..... And my system is keeping switching route for each data sent ..... and cause disconections or the IM network is impossible to connect ...... Is there a way to force a specific route depending on the destination ports ? An iptable rule missing ? an ip route setting missing ? don't know Thanks for your help ! Here is my ip route list (I've put some X for security reason): 82.234.XXX.0/24 dev eth0 proto kernel scope link src 82.234.XXX.XXX 192.168.0.0/24 dev eth1 proto kernel scope link src 192.168.0.254 82.228.XXX.0/24 dev eth2 proto kernel scope link src 82.228.XXX.XXX default equalize nexthop via 82.234.XXX.XXX dev eth0 weight 1 nexthop via 82.228.XXX.XXX dev eth2 weight 1 Here is my iptable rules : Chain INPUT (policy DROP) target prot opt source destination INETIN all -- anywhere anywhere ACCEPT all -- 192.168.0.0/24 anywhere ACCEPT all -- anywhere anywhere Chain FORWARD (policy DROP) target prot opt source destination INETIN all -- anywhere anywhere INETOUT all -- anywhere anywhere ACCEPT all -- 192.168.0.0/24 anywhere Chain OUTPUT (policy ACCEPT) target prot opt source destination INETOUT all -- anywhere anywhere Chain INETIN (2 references) target prot opt source destination ACCEPT icmp -- anywhere anywhere icmp echo-request ACCEPT icmp -- anywhere anywhere icmp echo-reply ACCEPT all -- anywhere anywhere state ESTABLISHED ACCEPT udp -- anywhere anywhere udp dpts:1024:65535 ACCEPT tcp -- anywhere anywhere tcp dpts:1024:65535 Chain INETOUT (2 references) target prot opt source destination ACCEPT all -- anywhere anywhere Chain LOG_ACCEPT (0 references) target prot opt source destination LOG all -- anywhere anywhere LOG level warning ACCEPT all -- anywhere anywhere Chain LOG_DROP (0 references) target prot opt source destination LOG all -- anywhere anywhere LOG level warning DROP all -- anywhere anywhere |
|
|||
|
On Sat, 26 Feb 2005 03:20:44 +0100, Judicator <judicator@ifrance.com>
wrote: >Hello, > >I have a load balanced linux router working (thanks to lartc website). > >In fact, It's working too much well ..... Some protocols, like IM >protocols (ICQ, MSN) need to keep the same route with the server ..... >And my system is keeping switching route for each data sent ..... and >cause disconections or the IM network is impossible to connect ...... > > >Is there a way to force a specific route depending on the destination >ports ? > >An iptable rule missing ? an ip route setting missing ? don't know Did you apply Julian's route patch? It is not just for Dead Gateway Detection. http://www.ssi.bg/~ja/dgd.txt http://www.ssi.bg/~ja/dgd-usage.txt http://www.ssi.bg/~ja/nano.txt Have you looked at this? Perhaps there is something common that could be used in applying CONNMARK. BEWARE: nexthop will probably not compile. http://selab.edu.ms/twiki/bin/view/N...inuxNetworking I have just the opposite "problem"; `ip route flush cache` is required to shake loose the persistent routing. http://linux-ip.net/html/tools-ip-ro...te-flush-cache Here is how I special case by IP ip rule add prio 50 to $DestIP/32 table $TBL where $TBL points to the interface I want the connection to use. buck |
|
|||
|
buck a écrit :
> On Sat, 26 Feb 2005 03:20:44 +0100, Judicator <judicator@ifrance.com> > wrote: > > >>Hello, >> >>I have a load balanced linux router working (thanks to lartc website). >> >>In fact, It's working too much well ..... Some protocols, like IM >>protocols (ICQ, MSN) need to keep the same route with the server ..... >>And my system is keeping switching route for each data sent ..... and >>cause disconections or the IM network is impossible to connect ...... >> >> >>Is there a way to force a specific route depending on the destination >>ports ? >> >>An iptable rule missing ? an ip route setting missing ? don't know > > > Did you apply Julian's route patch? It is not just for Dead Gateway > Detection. > http://www.ssi.bg/~ja/dgd.txt > http://www.ssi.bg/~ja/dgd-usage.txt > http://www.ssi.bg/~ja/nano.txt > > Have you looked at this? Perhaps there is something common that could > be used in applying CONNMARK. BEWARE: nexthop will probably not > compile. > http://selab.edu.ms/twiki/bin/view/N...inuxNetworking > > I have just the opposite "problem"; `ip route flush cache` is required > to shake loose the persistent routing. > http://linux-ip.net/html/tools-ip-ro...te-flush-cache > > Here is how I special case by IP > ip rule add prio 50 to $DestIP/32 table $TBL > where $TBL points to the interface I want the connection to use. > > buck > hi, I'll first try to apply the julian patch ...... I thought that it was only for helping things :-) thanks a lot for your help |
|
|||
|
About your trouble :
have a look to the netsane script : http://muse.linuxmafia.org/netsane/ maybe It'll work with it ;-) buck a écrit : > On Sat, 26 Feb 2005 03:20:44 +0100, Judicator <judicator@ifrance.com> > wrote: > > >>Hello, >> >>I have a load balanced linux router working (thanks to lartc website). >> >>In fact, It's working too much well ..... Some protocols, like IM >>protocols (ICQ, MSN) need to keep the same route with the server ..... >>And my system is keeping switching route for each data sent ..... and >>cause disconections or the IM network is impossible to connect ...... >> >> >>Is there a way to force a specific route depending on the destination >>ports ? >> >>An iptable rule missing ? an ip route setting missing ? don't know > > > Did you apply Julian's route patch? It is not just for Dead Gateway > Detection. > http://www.ssi.bg/~ja/dgd.txt > http://www.ssi.bg/~ja/dgd-usage.txt > http://www.ssi.bg/~ja/nano.txt > > Have you looked at this? Perhaps there is something common that could > be used in applying CONNMARK. BEWARE: nexthop will probably not > compile. > http://selab.edu.ms/twiki/bin/view/N...inuxNetworking > > I have just the opposite "problem"; `ip route flush cache` is required > to shake loose the persistent routing. > http://linux-ip.net/html/tools-ip-ro...te-flush-cache > > Here is how I special case by IP > ip rule add prio 50 to $DestIP/32 table $TBL > where $TBL points to the interface I want the connection to use. > > buck > |