Hi,
On Sonntag 16 März 2008, Freifunk Dresden wrote:
Hi,
I have found that the batmand-experimental (rev1003) still consumes more and more memory if you have specified the -r1 or -r2 options. I haven't tried -r3. If I turn off gateway with option -r0 the memory consumtion is constant. It seems that the memory increase is in steps of 16kbyte but the time when this happens is different.
the syslog does not show any hint despite of several tries to name the bat0 interface (about ever two minutes): Mar 16 14:50:49 (none) kern.err bmxd[2133]: Startup parameters: /sbin/batmand -s 10.12.0.1 -a 10.12.10.16/28 -r 1 --t 63 --no-unreachable-rule --no-throw-rules --no-prio-rules --one-way-tunnel 1 --two-way-tunnel 0 eth1 tbb /t 1 /i /A Mar 16 14:52:27 (none) kern.err bmxd[2398]: Trying to name tunnel to bat0 ... Mar 16 14:52:27 (none) kern.err bmxd[2398]: success! Mar 16 14:54:04 (none) kern.err bmxd[2601]: Trying to name tunnel to bat0 ... Mar 16 14:54:04 (none) kern.err bmxd[2601]: success!
Can you give more information about the scenario that might cause the problem and send a debug-level 3 log of the scenario. Maybe I got an idea then.
Both interfaces eth1 and tbb are active and conntected. The WRT54 that has the option -r0 set, has a constant memory consumtion.
During this test I got a very strange message on the wrt that only has the eth1 interface. There is no other router with the same ip address.
Mar 16 14:52:21 (none) kern.err bmxd[21119]: Drop packet: DAD alert! OGM from 10.12.10.17 via NB 10.12.10.17 with out of range seqno! rcvd sqno 28347, last valid seqno: 23290 at 6096067! Maybe two nodes are using this IP!? Waiting 0 more seconds before r
This DAD (duplicate address detection) message indicates a _potential_ duplicate address usage. It is triggered by receiving an invalid (or out-of-range) sequence number from the same IP address. During the start of each daemon an initial sequencenumber is randomized and then incremented by one with each new emitted OGM (by default with an originator interval of one second). A receiving node can now detect abnormal sequences. E.g. after receiving the sequence number n from node A, the next expected sequence numbers should be something around n+1. Ten seconds after receiving sequence number n+1 a sequence number around n+11 would be expected. This works even if some OGMs (and corresponding sequence numbers got lost). If instead of n+11 a sequence number of n+5000 is received then something is strange. Either two nodes are using the same IP and started with two different initial sequence numbers or only one node is using this IP address but the daemon on that node has been restarted. Then a new initial sequence number is randomized and other nodes will be temporary confused. The latter is probably the scenario indicated by your syslog. It logs: - an "out-of-range sequence number has been detected - tells you that related OGMs will be temporary ignored - and for how long it will continue to ignore these strange sequence numbers. If no more in-range sequence numbers arrive for a certain time the data set of node A is reinitialized and subsequent OGMs from node A will be accepted.
If you look at the debug-output of 10.12.10.17 it should indicate that this node has been restarted and thereby changed its sequence number counter from 23290 to 28347.
Hope the explanation helped bit.
regards, axel
Setup:
laptop-tbb-----------tbb[10.12.0.17]eth1--------eth1[10.12.10.1]
+------------tbb[10.12.10.17]eth1---------+/stephan
B.A.T.M.A.N mailing list B.A.T.M.A.N@open-mesh.net https://list.open-mesh.net/mm/listinfo/b.a.t.m.a.n