12-20-2017 12:06 AM
12-20-2017 12:45 AM - edited 12-20-2017 12:46 AM
Yes looking like storage port is not able to handle frames from the servers.
What are the port speed of the servers and storage?
This might be a Q depth issue, an issue on storage itself or a wrong optimization on it.
12-20-2017 04:17 AM
12-20-2017 07:06 AM
Troubleshooting class 3 discards, and device latency issues is a large discussion and can include many aspects of a SAN. A slow draining device can cause this, a poor performing HBA, or a fan ratio issue can contribute, and ISL oversubscription can also affect latency. Please note that port 18 in the report is the port which is BEING affected, and is not likely the port which is the cause of the problem.
First things to do: run the commands: statsclear; slotstatsclear on each switch in the fabric. After 24-48 hours run porterrshow on each switch in the fabric. Review those outputs for errors, and advise what you find. Also provide your output of firmwareshow, fabricshow, so that we know what kind of equipment we are looking at.
I would advise you to run the SAN Health report, and gather some info on your connections, and througput. Without a complete picture of the fabric, and all F and E port connections it will be impossible to diagnose. There are also records in the log file which may be useful in determining what ports are causing latency within the fabric.
You will get a report emailed to you showing the switch connections, and many attributes of the fabric. Once that report is checked, we can proceed with some options. As a first guess, try to find legacy devices which may be running at 4GB and are traversing the fabric via ISL(E_ports). This is a common issue causing class 3 discards, but it is only one possible issue. There are many other things which affect throughput and congestion.