I havent seen that on my RB1000. Do you have the ports locked down to a set rate?
On Wed, May 26, 2010 at 1:16 PM, Kevin Sullivan <[email protected]> wrote: > I think I have found a legitimate bug. I'm running an RB1000 that > we put in service about 2 weeks ago (it replaced another RB1000 that was > having similar problems). Here is what is going on: > > Linux router "A" <---------> [ether1] RB1000 [ether3] <-----------> Linux > router "B" > > > The RB1000 above is connected to the two hosts shown. > Each link A < > RB1000 < > B has latency ~1ms. We are not using any > Mikrotik wireless. > > A and B both know that they can reach each other through the RB1000 (thanks > to OSPF). > > A and B are Linux routers. When I ping B from A (traffic going through the > RB1000), I get no response. When I log into B and tcpdump traffic, I can > see icmp echo request packets coming in from A, and echo reply packets going > out to A. Fine. I then log into the RB1000 and packet sniff ether1 and > ether3. > > ether1 packet sniff shows icmp request packets coming in. ether3 shows icmp > request packets going out, and icmp reply packets coming in. However, the > replies are not going out ether1. > > BUT.... after several minutes, A starts seeing replies. > > 64 bytes from x.x.x.x: icmp_seq=1049 ttl=63 time=671216 ms > 64 bytes from x.x.x.x: icmp_seq=1095 ttl=63 time=628217 ms > 64 bytes from x.x.x.x: icmp_seq=1142 ttl=63 time=584217 ms > 64 bytes from x.x.x.x: icmp_seq=1188 ttl=63 time=541218 ms > 64 bytes from x.x.x.x: icmp_seq=1235 ttl=63 time=497234 ms > > the RB1000 has been queuing my ICMP packets for ~500 seconds!! > > I STOP pinging from A and packet sniff ether1 on the RB1000 again. It is > STILL sending out queued ICMP replies from A, even though I am > not sending requests anymore. > > Several minutes after I stop pinging from A, the RB1000 stops sending > replies on ether1. > > Clients have been complaining for months about slow speeds passing traffic > through this router. I've also noticed high CPU utilization, even when > normal CPU hungry tasks were turned off (one mangle rule, no queues, no > proxy, no DNS, etc). During the day, we see 70-80% CPU utilization. The > previous router (same config) went to 100% utilization, which is why we > replaced it. > > Regards, > > Kevin > > > > > > > > -------------------------------------------------------------------------------- > WISPA Wants You! Join today! > http://signup.wispa.org/ > -------------------------------------------------------------------------------- > > WISPA Wireless List: [email protected] > > Subscribe/Unsubscribe: > http://lists.wispa.org/mailman/listinfo/wireless > > Archives: http://lists.wispa.org/pipermail/wireless/ > -------------------------------------------------------------------------------- WISPA Wants You! Join today! http://signup.wispa.org/ -------------------------------------------------------------------------------- WISPA Wireless List: [email protected] Subscribe/Unsubscribe: http://lists.wispa.org/mailman/listinfo/wireless Archives: http://lists.wispa.org/pipermail/wireless/
