Unsolved
This post is more than 5 years old
2 Posts
0
21606
March 1st, 2007 11:00
icmp redirects from 6248
Sorry, I didn't realize that the post would get mangled. If there is a moderator here, please remove my previos post.
My problem is that I randomly get strange/incorrect icmp redirects from my switches. I have the following setup:
-Two 6248's interconnected using a trunk LAG with 2 ports.
-VLAN3 configured on both switches and tagged over the LAG.
-Both switches have IPs on the vlan, 10.0.3.11 and 10.0.3.12
-A host, HostA 10.0.3.101, connected to a port on VLAN3 on switchA.
HostA is running Nagios and pings the two switches on regular intervals. Everything works fine for a while until HostA suddenly starts getting duplicate icmp replies combined with icmp redirect replies when pinging switchA. A tcpdump on HostA shows:
10:06:15.104709 IP 10.0.3.101 > 10.0.3.11: icmp 64: echo request seq 0
10:06:15.105576 IP 10.0.3.11 > 10.0.3.101: icmp 64: echo reply seq 0
10:06:15.105951 IP 10.0.3.12 > 10.0.3.101: icmp 36: redirect 10.0.3.11 to host 10.0.3.11
10:06:15.106451 IP 10.0.3.11 > 10.0.3.101: icmp 64: echo reply seq 0
Thus, the packets sent to 10.0.3.11 somehow "leaks" to SwitchB (who behaves correctly in sending the icmp redirects back to HostA). Meanwhile, HostA have no problems reaching other hosts connected to either switch. It's only packages addressed to the switch itself that causes problems. Then after a while it all turns back to normal.
I can also see the problem when pinging SwitchB from HostA through SwitchA. Then the problem is that SwitchA seems to catch packets that it should just forward.
Just switch IPs .11 and .12 in the tcpdump above to see how that looks.
Anyone have any ideas why this is occuring and what can be done to fix it? Bug in the switch or in my config? I can post the config if needed.
A fix would be to disable the sending of icmp redirects, as can be done on Cisco, but I see no way of doing this on the 6248. Can it be done?
My problem is that I randomly get strange/incorrect icmp redirects from my switches. I have the following setup:
-Two 6248's interconnected using a trunk LAG with 2 ports.
-VLAN3 configured on both switches and tagged over the LAG.
-Both switches have IPs on the vlan, 10.0.3.11 and 10.0.3.12
-A host, HostA 10.0.3.101, connected to a port on VLAN3 on switchA.
HostA is running Nagios and pings the two switches on regular intervals. Everything works fine for a while until HostA suddenly starts getting duplicate icmp replies combined with icmp redirect replies when pinging switchA. A tcpdump on HostA shows:
10:06:15.104709 IP 10.0.3.101 > 10.0.3.11: icmp 64: echo request seq 0
10:06:15.105576 IP 10.0.3.11 > 10.0.3.101: icmp 64: echo reply seq 0
10:06:15.105951 IP 10.0.3.12 > 10.0.3.101: icmp 36: redirect 10.0.3.11 to host 10.0.3.11
10:06:15.106451 IP 10.0.3.11 > 10.0.3.101: icmp 64: echo reply seq 0
Thus, the packets sent to 10.0.3.11 somehow "leaks" to SwitchB (who behaves correctly in sending the icmp redirects back to HostA). Meanwhile, HostA have no problems reaching other hosts connected to either switch. It's only packages addressed to the switch itself that causes problems. Then after a while it all turns back to normal.
I can also see the problem when pinging SwitchB from HostA through SwitchA. Then the problem is that SwitchA seems to catch packets that it should just forward.
Just switch IPs .11 and .12 in the tcpdump above to see how that looks.
Anyone have any ideas why this is occuring and what can be done to fix it? Bug in the switch or in my config? I can post the config if needed.
A fix would be to disable the sending of icmp redirects, as can be done on Cisco, but I see no way of doing this on the 6248. Can it be done?
No Events found!