1

We presume to have a faulty cable that connects the SAN to a direct I/O LDOM. This is a snippet of the error when running iostat -En

   c5t60060E8007C50E000030C50E00001067d0 Soft Errors: 0 Hard Errors: 696633 Transport Errors: 704386
Vendor: HITACHI  Product: OPEN-V           Revision: 8001 Serial No: 504463
Size: 214.75GB <214748364800 bytes>
Media Error: 0 Device Not Ready: 0 No Device: 6 Recoverable: 0
Illegal Request: 1 Predictive Failure Analysis: 0

What does No Device: 6 mean here?

2
  • What HBA/device driver are you using? Commented Mar 5, 2019 at 23:07
  • HBA Port WWN: 21000024ff470f2e OS Device Name: /dev/cfg/c2 Manufacturer: QLogic Corp. Model: 371-4522-02 Firmware Version: 8.05.00 FCode/BIOS Version: BIOS: 2.10; fcode: 3.06; EFI: 2.04; Serial Number: 0402T00-1232085736 Driver Name: qlc Driver Version: 170516-5.06c Type: N-port State: online Supported Speeds: 2Gb 4Gb 8Gb Current Speed: 8Gb Node WWN: 20000024ff470f2e Link Error Statistics: Link Failure Count: 4386
    – hovey
    Commented Mar 6, 2019 at 1:18

1 Answer 1

0

A search through the Illumos fiber-channel device code for ENODEV shows 13 uses of ENODEV in the source code that originated as OpenSolaris.

Of those instances, I suspect this is the one most likely to cause your "No device" errors:

pd = fctl_hold_remote_port_by_pwwn(port, &pwwn);
if (pd == NULL) {
    fcio->fcio_errno = FC_BADDEV;
    return (ENODEV);
}

That code is in the fp_fcio_login() function, where the code appears to be trying to login to a remote WWN. It seems appropriate to assume a bad cable could prevent that from happening. Note that fiber-channel error code is FC_BADDEV, which also seems appropriate for a bad cable.

In short, a review of the source code indicates that ENODEV errors are consistent a bad cable.

You can use dTrace to more closely identify the association if necessary. Given that both hard and transport errors occur about 5 or 6 orders of magnitude more frequently, IMO that effort isn't necessary until the ENODEV errors occur after the other errors are addressed and no longer occur.

2
  • Thanks. So is 'No Device' a counter of sorts?
    – hovey
    Commented Mar 6, 2019 at 20:30
  • In this case, the number you are seeing is a count of the number of times that error code was returned when something didn't work. But if you look at your total number of errors, you have 6 "No device" errors likely generated when your system had trouble connecting to a remote disk controller/LUN compared to 1.4 million hard and transfer errors. Commented Mar 8, 2019 at 13:42

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .