TX Wait troubleshooting

We have some SX6012 40Gb ethernet switches that connect a several ESXi servers.

We are seeing some latency with our iSCSI SAN storage that looks to be in the network stack. During troubleshooting we noticed that the ESXi switch ports have a high count of TX Wait.

If we reset the counters we will have a few hundred counts for TX Wait within a few seconds.

Doucmentation says this can be due to " insufficient credits or because of lack of arbitration" but I’m not really sure what that means. These switches only have 3 of the 12 ports in use and really shouldn’t be saturating the switch.

What are next steps to figure out what is affecting the TX Wait and/or if thats related to our latency issues.