I have a datacenter config of 2 x 7k Nexus connected via vPC, 2 x 5k connected via vPC, 2x 2232 FEX connected to 5ks and 8 x 2248 FEX connected to 5ks.
I am experiencing intermittent network loss on systems connected on the FEXs where the symptoms displayed are:
1. A normal 32 byte ping replies with no errors.
2. A loaded ping with a size greater than 1472 results in all packets being dropped.
3. Not all systems are effected at the same time.
4. If I change the IP of an effected system the issue resolves itself.
5. If I set the IP of a another system to the previously not working system IP that system does not display any issues.
6. All the systems connected to the Nexus environment are running windows server 2012.
7. The server hardware is HP proline G8s and G7s.
8. All systems connected to the old Catalyst 6500 environment do not have the issues.
We went through the process of updating the 5ks to the latest firmware and therefore all the FEXs as well.
We had previously updated the HP NIC drivers to the latest version and decided to roll back to a previous version.
We have enabled the jumbo frames in the policies on the 7Ks as well as the 5ks.
All of the above actions resolved the issue for 4 days and then all the systems starting presenting again and now I am at a loss to understand what is happening.
Ca anyone tell me if they have experienced this issue before and if there is a resolution?