recently i have combined 2 machines into 1. after doing this, the ping to the new machine stops, for no apparent reason, and consequently all traffic to/from the machine also stops. when using the two machines individually this never happened.
one time the ping goes like this :
Code:
Reply from 10.10.1.254: bytes=32 time<1ms TTL=64
Reply from 10.10.1.254: bytes=32 time<1ms TTL=64
Reply from 10.10.1.19: Destination host unreachable.
Reply from 10.10.1.19: Destination host unreachable.
Reply from 10.10.1.19: Destination host unreachable.
Reply from 10.10.1.19: Destination host unreachable.
Reply from 10.10.1.19: Destination host unreachable.
Reply from 10.10.1.254: bytes=32 time=999ms TTL=64
Reply from 10.10.1.254: bytes=32 time<1ms TTL=64
Reply from 10.10.1.254: bytes=32 time<1ms TTL=64
another time it wont drop the packet, but the ping is huge...
Code:
Reply from 10.10.1.254: bytes=32 time<1ms TTL=64
Reply from 10.10.1.254: bytes=32 time<1ms TTL=64
Reply from 10.10.1.254: bytes=32 time=1001ms TTL=64
Reply from 10.10.1.254: bytes=32 time<1ms TTL=64
Reply from 10.10.1.254: bytes=32 time<1ms TTL=64
also after a long time, (2-3 days) i get 'neighbor table overflow' listed in the console of the machine. i have increased the gc_timeout or whatever, and this does supress the warning, but my problem is still there. the neighbor table overflow, in my understanding is only a problem with a large number of hosts (for default sysctl parameters).
Lastly, if i restart the network, it kills everything (like the network just wont restart - it reaches the last interface and hangs), and if i reboot the machine, it seems to be fine for about 3-4hours.
has anyone seen anything like this when connecting a large number of hosts to 1 gateway, using linux as the router?