Lately I have had the pleasure of discovering a new issue that has risen up from the 1s and 0s. Currently, there is a network that is not using LACP but using NIC teaming, this is fine. The weird part is, one of their UC VMs randomly stopped talking outside the subnet. Rebooting the VM did nothing and it kept looking like it was more and more of a network problem. A TAC case was opened and even Cisco was pointing the finger to the network.
Without being to in detail on ESXi, since I'm no guru, I took a look at the network side of the house and everything looked like it was in working order as well. Pings to and from the ESXi were working but for some strange reason on VM was sending data out one NIC and receiving it on the other. This was the only VM doing this and I think if LACP was enabled it would never have happened. The end result was a VM that wasn't talking correctly and couldn't communicate with it's subscriber. We ended up moving one of the NICs from active to standby and ran a test. Everything magically worked itself out as all traffic was being forced over one lane. We then put it back into active mode and everything still was working fine. Weird.... How did the TCP/UDP data streams end up getting hosed up?
I checked again today and traffic was still evenly distributed and no weirdness was going on behind the scenes with one VM sending out one port and receiving on the other. I'm not sure this was ever a problem as again, I'm not an ESXi expert but it seemed to have resolved the issues and they haven't returned. This also seemed to have fixed a second issue they were getting which was calls going to voicemail but then terminating after 5 seconds. If this was indeed the resolution, nothing was being shown as wrong anywhere on the network and no errors were being thrown. Just a heads up for any of you that run into this issue yourselves. Just bounce the port or go standby then back to active, it should end with the same result.
No comments:
Post a Comment