We see lot's packet drops on our cisco ip sla probes
cishoejnielsen opened this issue · 16 comments
Hi,
We have app. 5000 probes pointing to the responder. We are seeing lots of packet loss.
It's a centos 5.6
ifconfig on eth0 is 195.41.248.165.
the responder config is very simple:
vlansupport=0
iface=eth0
debuglevel=0
ciscoport=64000
We don't see any drops on eth0 , and cpu and memory load is very little
Can you please help
Regards
Carsten
Where are you seeing the packet loss? Can you describe your setup?
Can you verify that the responder address is not mapped to any interface on the responder machine and that you are routing the packages to the responder machine's interface IP. This might sound bit illogical, but the responder software is supposed to care of it's own IP package handling.
the IP address must not be configured on any interface on the responder host. the responder software will handle it's own IP stack. you must route the responder service IP towards the responder host IP.
Github seems to ignore attachments over email.
Also, have you checked that your CoPP filter does not ratelimit packets?
thank you. can you confirm with packet sniffing on egress point that the server is indeed receiving more packets than it's not responding to? it also seems you are using virtual environment for running the software, and it might be the cause of your trouble. Last time I tested this with virtual machine it had trouble working correctly due to limitations in virtual platform.
sorry to jump on this after two years, but could i know what was the VM configuration and the physical server configuration you used?. i'm planning to test this, but most probably will not have option of physical server. I don't think there will be more than 2000 probes per vm.