Qla2xxx Kernel Recovery Tips

Resolution

• Contact your storage vendor for help finding a solution.
  • Repeated and consecutive attempts to regain access to the device fail. Thus, the remote device is considered autonomous.
• An error occurred while gardening on the SAN. A workaround in some failure cases was to exclude automatic connection and channel speed and set the FC speed to a fixed value of 4 GB instead of 8 GB.
• Also check hardware caused by removing / disconnecting storage devices from a new server that is not in use.

Reason

 Dec 15: 46: 30 8 Core: qla2xxx 0000: 06: 00.0: scsi (0: 10: 0): Reject the supplied command - 1 2002 10fb. [2] multiple dec 15:46:41 Kernel: qla2xxx 0000: 06: 00.0: scsi (0: 10: 0): Undo the issued command specific to 2002 10fb. [3]actions December 15:46:41 Kernel: qla2xxx 0000: 06: 00.0: scsi (0: 10: 0): DEVICE RESET DONE. [3]actions Dec 15:46:41 Core: qla2xxx 0000: 06: 00.0: scsi (0: 10: 0): SUCCESSFULLY RESET DEVICE. [3]Dec number 15:46:52 Core: qla2xxx 0000: 06: 00.0: scsi (0: 10: 0): Cancel command issued - obviously 2002 10fb. [5]11.12. 15:46:52 Kernel: qla2xxx 0000: 06: 00.0: scsi (0: 10: 0): CYCLE RESET DONE. [5]Dec 5 15:46:52 Kernel: qla2xxx 0000: 06: 00.0: LOOP DOWN detected 0 (0 0). [5]December 8 15:46:57 Core: qla2xxx 0000: 06: 00.0: LOOP Upward motion detection (8 Gbps). [5]December 10 15:47:18 Core: qla2xxx 0000: 06: 00.0: qla2xxx_eh_bus_reset: Adaptation was successful [5]December 2 15:47:39 Core: 0000: 06: 00 qla2xxx.0: scsi (0: 10: 0): Undo the issued command - 1 2004 10fb. [6]December number 15:47:39 Idraw: 0000: 06: 00 qla2xxx.0: scsi (0: 10: 0): ADAPTER RESET DONE. [6]Dec 8 15:47:39 Core: 0000: 06: 00 qla2xxx.0: Execute provider medical error - ffff810139dd44f8 ha =. [6]Dec 8 15:47:45 Kernel: qla2xxx 0000: 06: 00.0: LOOP detected at full height (8 Gbps). [6]5.Dec. 15:47:47 qla2xxx Core: 0000: 06: 00.0: qla2xxx_eh_host_reset: Adaptation was successful [6]December 10, 15:48:08 Core: qla2xxx 0000: 06: 00.0: scsi (0: 10: 0): Undo the issued command - 1 2005 10fb. [7]Dec 8-10 15: 48: 08 Kernel: st 0: 0: 10: 0: scsi: device offline - not ready after troubleshooting [7]December 8, 15:48:08 Core: st 0: 0: 10: 0:  Time and command, expected 14000 sec [1] 

• [1] Some abortions are caused by delay; the exact reason for the built-in error handling is displayed only if all the work on repetitions and recovery exchanges for the device in question no longer works. Which was the case in this single case, so the categorical cause of the problem is a hardware timeout of 14000 seconds only for the I / O command for that particular device.
• [2] The first level of error handling is to simply rollback our own io and repeat the priestsweave.
• [3] The duplicate io no longer works, so it is canceled and the second step of the error correction reset is attempted: device. The device reset will complete successfully and the current I / O will be repeated (sent to the hardware).
• [4] Retry failed again. The third step recovery error is to reset the save target. Not all storage HBAs support nearly all of the target level reset control functions and will be ignored if they are not supported, as in this case.
• [5] Since the re-I / O failed and the next reset step) (target unreachable, my fourth troubleshooting step is actually done: bus reset (LOOP RESET in this case is the same as bus reset) – tourist bus connection is interrupted and reconnecting, mostly because it can be seen through the downstream / upstream loop. Bus reset succeeded and an additional attempt will be made from io (sent to hardware for you).
• [6] Retry will fail if. The fourth and final step is performed, relatednd with step failure recovery: reset the adapter to zero. If it is professional, then io will be repeated more often for now (send to equipment).
• [7] Repeated io doesn’t work again. After all attempts to repair the hardware have been exhausted and the I / O type is still not working correctly, the device will not be connected to the network and the cause of the failure will be displayed. In this state, the main reason is a timeout.

Here are 5 standard steps to escalate Scsi troubleshooting initiatives:

1. Cancel, please try again
2. Drop the device and reset, try again
3. Drop and reset target storage, try again
4. Exit and recreate the bus, please try again
5. Cancel and reset adapter settings, try again.

Thus, after five attempts and a series of attempts to restore successful communication in order to restore the device in question, all attempts to maintain it have failed. At the moment, each of our systems has little other choice, but they can disable the deviceto avoid such attempts in future I / O operations. Restricting future attempts prevents other devices on the same bus from disrupting operation when the bus and adapter are rebooted, and affects all devices on the same HBA / transport.

Diagnostic Steps

• This can indicate a number of dilemmas: HBA, SAN, or ST-band. Check SAN connectivity as storage.