Is this for exam purposes, or actual problem?
For real world troubleshooting,,, there are a number of places the issues can exist including the OS, config.
Generally best NOT to enable jumbo frames until after you have general connectivity and your configs ironed out.
Target and initiators should always be L2 adjacent, not go through any L3 gateways, and use as few port hops as possible.
Make sure your OS NIC drivers are up to date, and that you don't see any CRC counters incrementing anywhere along the path between initiator and target.
For some storage vendor iSCSI solutions, QOS related configs also come into play.
It is a real problem, collegues are having problem regarding ESXI hosts. So I first need to look into coonectivity from ESXI to Nexus(2k FEX). Its a two tier nexus data center design.
FEXs are generally not recommended for heavy storage traffic (for instance we don't support those for use in Hyperflex).
Also if you have hosts on the same fex that are transmitting a lot of traffic to each other, that traffic has to go back up to your parent switch to be 'switched', which makes it much easier to saturate the fex-to-parent switch uplinks than you might initially think.
Some times a quick wireshark session can show lots of retransmits if you have tail drop style of issues.
Take a look at https://www.cisco.com/c/en/us/support/docs/switches/nexus-2000-series-fabric-extenders/200265-Troubleshooting-Fabric-Extender-FEX-Pe.html for some good troubleshooting on FEX related.