Random Network Slowness [message #40820] |
Wed, 06 October 2010 21:08 |
cbschuld
Messages: 5 Registered: April 2009 Location: Phoenix, AZ
|
Junior Member |
|
|
Hello All:
I have a physical HN with four bonded NICs bonded using mode balance-alb
I am using venet and have a single venet0 instance
I have four CTs running on this single HN. The HN is a Dell PowerEdge R610 and the CTs are under low stress.
Everything works great; however, I am experiencing random timeouts from different services such as CT1 has MySQL on it and CT2 is connecting to it. Occasionally I will get a timeout on CT1 from CT2.
After studying this for some time I realized that from CT1 if I ping CT2 using CT2's IP address I see something like this
.
.
.
.
64 bytes from 10.0.0.13: icmp_seq=171 ttl=63 time=0.138 ms
64 bytes from 10.0.0.13: icmp_seq=172 ttl=63 time=0.194 ms
64 bytes from 10.0.0.13: icmp_seq=173 ttl=63 time=0.143 ms
64 bytes from 10.0.0.13: icmp_seq=174 ttl=63 time=0.148 ms
64 bytes from 10.0.0.13: icmp_seq=175 ttl=63 time=0.205 ms
64 bytes from 10.0.0.13: icmp_seq=176 ttl=63 time=0.203 ms
64 bytes from 10.0.0.13: icmp_seq=177 ttl=63 time=0.194 ms
64 bytes from 10.0.0.13: icmp_seq=178 ttl=63 time=0.170 ms
64 bytes from 10.0.0.13: icmp_seq=180 ttl=63 time=0.194 ms
64 bytes from 10.0.0.13: icmp_seq=181 ttl=63 time=0.191 ms
64 bytes from 10.0.0.13: icmp_seq=182 ttl=63 time=0.212 ms
64 bytes from 10.0.0.13: icmp_seq=183 ttl=63 time=500 ms
64 bytes from 10.0.0.13: icmp_seq=184 ttl=63 time=0.172 ms
64 bytes from 10.0.0.13: icmp_seq=185 ttl=63 time=0.126 ms
64 bytes from 10.0.0.13: icmp_seq=186 ttl=63 time=0.145 ms
.
.
.
.
The ping/connectivity/response is fine but about every 30s - 60s I see one ICMP packet have a HUGE delay on it. This is VERY similar to what my applications are experiencing as well --> works fine, works fine, works fine, TIMEOUT, works fine, works fine, works fine, etc...
Additionally, this is ONLY relative to the machines using the venet0 interface. If I use an external machine to the HN and ping or use any of the services that are on the machines hooked to the venet0 I receive no timeouts. Everything works great. It is only if one venet0 connected CT connects to another venet0 connected CT. Furthermore, I have several of these HNs and it also effects any CT on any of those machines. So this occurs if any venet0 mounted CT attempts to use services/ping from another CT.
Has anyone experienced this or have any ideas?
Thanks, Chris
[Updated on: Wed, 06 October 2010 21:25] Report message to a moderator
|
|
|