FC traffic stops 2 min and resume. QMI8142 - NX 4001 - NX5548

[Guido Arturo Catalano]

Hi, I have an IBM Blade H with 2 Nexus 4000, bladeservers HS23 with Qlogic QMI8142. FCF are two Nexus 5548UP, 6.0(2)N1(1).

VSAN VLAN from each NX4k are connected to only one NX5k using trunk with only VSAN and VLAN 1 allowed. (I don't know why, but without VLAN 1, the bladeserver's can't access FCF)

* FC Tape function is disabled on QLogic BIOS

* Every server has independent zone with each storage

* Each NX4k/5k pair, are in different VSANs

* Multipath is disabled during windows setup

* Every firmware are up to date. (NX 4k, 5k, QLogic, Chasis, Bladeservers, Storage, disks)

Everything looks ok, uEFI and windows installer saw LUN 0. But, windows setup took up to 12hs. Looking at storage ( I tried with a DS3500 and a N3400, both FC, attached to one or both NX5k) I see normal use for a while and then nothing at all. 30 seconds at 300 IO/s and then, 4 minutes of 0 IO/s.

Windows wait without complains and eventually finish and boots ok. Inside windows, I've tested differents drivers but always the same thing happend.

With windows performance meter on screen, I start a file copy. 200MB/s, 100MB/s and after a random (for me) amount of seconds, disk usage drops to 0KB/s. After some minutes, disk comes to life again and filecopy transfer continue without any warning.

Windows event log is perfect, despite C: stops responding for minutes, windows just freeze and continue as if nothing happened ever.

How can I analyse FC traffic to diagnose what is happening?

Thanks, I appreciate your help and time.

[Guido Arturo Catalano]

self update:

I've connected an external server (IBM x3650) using a QLE8142 to N4k and have the same issue. Then, I've connected the same server directly to N5k and it works fine!  So, the problem is related to N4k.

I'm going to review all the setup and start with basic connection to uplinks N5k.

My setup looks like this one:

http://www.cisco.com/en/US/prod/collateral/switches/ps9441/ps10596/deployment_guide_c07-574724.html#wp9000139

Guido

[Guido Arturo Catalano]

Solution:

After swapping with a new N4k a few times, we discover a problem in the midplane of the chassis. The problem was resolved pressing firmly on both NX4001.

Apparently, the switches were locked in position but don't work before pressing on the back of them. I guess that a faulty internal connector in the midplane cause the problem.

Guido.