Hi Axel,
when I use rv980 and call "kill -STOP $(pidof tincd)" then batman works for a while (1-2 minutes) and exists with the following error:
[ 411413] Error - can't send udp packet: Resource temporarily unavailable. Connection terminated by remote host
When I restart batmand when tincd is still stopped, the OGM are processed and I see the neighbour nodes until I get the same error after about 1-2 minutes. good news, it does not block.
Because of the error message I have inserted at posix/posix.c: 705 "return 0" (send_udp_packet()).
I have tested two wrt54gl that had the problems before with the rev980 and the "return 0" statement. Both routers has be running for 18 hours when tincd was running without creating a connection. After that I called "kill -STOP $(pidof tincd)" on both routers. Both routers running now since 13 hours without problems.
Please check and insert the "return 0" at posix/posix:705 if it is the right solution.
Have a nice week
/Stephan
Zitat von Axel Neumann axel@open-mesh.net:
Hi Stephan,
can you check if rv980 solves the problem ?
ciao, axel
On Donnerstag 14 Februar 2008, Freifunk Dresden wrote:
Hi Axel,
I have done some checks. First I have added the tincd tap interface to a bridge and used the bridge as parameter for batmand. This is currently running since 10h.
The second what I have tried is to call "kill -STOP $(pidof tincd)". On the WRT54GL batmand stopped quite fast, but it stopped also fast without the "kill". On the WRT54SL where I never have seen batmand stopping, the "kill" has stopped batmand after 5-10 minutes. I could reproduce this serveral times with the GS.
When calling batmand -c -a 120.56.56.56/32 I get the following output on batmand -cd3:
[ 275531] Unix socket: got connection [ 275539] got request: 10 [ 275540] Unix socket: Requesting adding of HNA 120.56.56.34/32 - put this on todo list... [ 275544] got request: 10 [ 275545] Unix client closed connection ...
Three batmand processes are running and batmand -c returns: WARNING: You are using BatMan-eXp 0.3-alpha (compatibility version 10) ! /sbin/batmand [not-all-options-displayed] -r 2 -a 10.12.10.16/28 eth1 tbb
After calling "killall batmand" only one of three batmand processes is killed. I have to killall -9 to get rid of it.
Regards /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
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
Hi Stephan,
thanks for that hint. I changed that lines with rv982 and also made the --resist-blocked-send switch the default behavior now. I also believe that terminating the daemon due to any temporary unavailable device or blocked port is too paranoid. This should also help with other previously reported problems like "[B.A.T.M.A.N.] strange batman aborting" indicated by Jan
ciao, axel
On Sonntag 17 Februar 2008, Freifunk Dresden wrote:
Hi Axel,
when I use rv980 and call "kill -STOP $(pidof tincd)" then batman works for a while (1-2 minutes) and exists with the following error:
[ 411413] Error - can't send udp packet: Resource temporarily unavailable. Connection terminated by remote host
When I restart batmand when tincd is still stopped, the OGM are processed and I see the neighbour nodes until I get the same error after about 1-2 minutes. good news, it does not block.
Because of the error message I have inserted at posix/posix.c: 705 "return 0" (send_udp_packet()).
I have tested two wrt54gl that had the problems before with the rev980 and the "return 0" statement. Both routers has be running for 18 hours when tincd was running without creating a connection. After that I called "kill -STOP $(pidof tincd)" on both routers. Both routers running now since 13 hours without problems.
Please check and insert the "return 0" at posix/posix:705 if it is the right solution.
Have a nice week
/Stephan
Zitat von Axel Neumann axel@open-mesh.net:
Hi Stephan,
can you check if rv980 solves the problem ?
ciao, axel
On Donnerstag 14 Februar 2008, Freifunk Dresden wrote:
Hi Axel,
I have done some checks. First I have added the tincd tap interface to a bridge and used the bridge as parameter for batmand. This is currently running since 10h.
The second what I have tried is to call "kill -STOP $(pidof tincd)". On the WRT54GL batmand stopped quite fast, but it stopped also fast without the "kill". On the WRT54SL where I never have seen batmand stopping, the "kill" has stopped batmand after 5-10 minutes. I could reproduce this serveral times with the GS.
When calling batmand -c -a 120.56.56.56/32 I get the following output on batmand -cd3:
[ 275531] Unix socket: got connection [ 275539] got request: 10 [ 275540] Unix socket: Requesting adding of HNA 120.56.56.34/32 - put this on todo list... [ 275544] got request: 10 [ 275545] Unix client closed connection ...
Three batmand processes are running and batmand -c returns: WARNING: You are using BatMan-eXp 0.3-alpha (compatibility version 10) ! /sbin/batmand [not-all-options-displayed] -r 2 -a 10.12.10.16/28 eth1 tbb
After calling "killall batmand" only one of three batmand processes is killed. I have to killall -9 to get rid of it.
Regards /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
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
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
b.a.t.m.a.n@lists.open-mesh.org
-
Axel Neumann -
Freifunk Dresden