SCSI Retry on BUSY

[chakri]

Hi,

observed the following scsi error messages on system. could some one explain me who prints this messages ( driver or SCSI fiber channel firmware) and what does task abort, target reset, bus reset means...

kernel: [0:0:0:0] SCSI Retry on BUSY
kernel: mptscsih: ioc0: attempting task abort! (sc=ffff8101303b1800)
kernel: sd 0:0:0:0:
kernel: command: cdb[0]=0x2a: 2a 00 00 07 68 9a 00 00 08 00
kernel: mptscsih: ioc0: task abort: SUCCESS (sc=ffff8101303b1800)
kernel: mptscsih: ioc0: attempting target reset! (sc=ffff8101303b1800)
kernel: sd 0:0:0:0:
kernel: command: cdb[0]=0x2a: 2a 00 00 07 68 9a 00 00 08 00
kernel: mptscsih: ioc0: Issue of TaskMgmt failed!
kernel: mptscsih: ioc0: target reset: FAILED (sc=ffff8101303b1800)
kernel: mptscsih: ioc0: attempting target reset! (sc=ffff8101303b19c0)
kernel: sd 0:0:0:1:
kernel: command: cdb[0]=0x2a: 2a 00 00 06 68 b0 00 00 18 00
kernel: mptscsih: ioc0: Issue of TaskMgmt failed!
kernel: mptscsih: ioc0: target reset: FAILED (sc=ffff8101303b19c0)
kernel: mptscsih: ioc0: attempting target reset! (sc=ffff8101193ce500)
watchdog[1363]: still alive after 464640 seconds = 46464 interval(s)
kernel: sd 0:0:0:2:
kernel: command: cdb[0]=0x2a: 2a 00 25 54 48 10 00 04 00 00
kernel: mptscsih: ioc0: Issue of TaskMgmt failed!
kernel: mptscsih: ioc0: target reset: FAILED (sc=ffff8101193ce500)
kernel: mptscsih: ioc0: attempting bus reset! (sc=ffff8101303b1800)
kernel: sd 0:0:0:0:
kernel: command: cdb[0]=0x2a: 2a 00 00 07 68 9a 00 00 08 00
kernel: mptscsih: ioc0: bus reset: SUCCESS (sc=ffff8101303b1800)

[Matti_Kurkela · Accepted solution]

All the messages prefixed by "kernel:" come from the kernel. As they're related to SCSI, they obviously come from some parts of the kernel's SCSI subsystem.

The messages prefixed by "kernel: mptscsih: " come from the SCSI driver that handles LSI Fusion MPT SCSI interfaces. The driver is a part of the SCSI subsystem, which is a part of the kernel. You can find these messages in the Linux kernel source, in the file
kernel-/drivers/message/fusion/mptscsih.c
if you're interested.

The message prefixed by "watchdog[1363]" is not related to the SCSI problem: it comes from the system watchdog process, which is running on PID 1363. It looks like an informational message, not an error.

Task abort: the driver says "stop whatever you're doing" to the SCSI disk (or whatever SCSI target device).

Target reset: the driver commands the SCSI disk to reset itself.

Bus reset: the driver resets the entire SCSI bus, causing *all* SCSI devices on that bus to reset. This operation has more force to clear temporary problems than a target reset, but it stops all data traffic on the bus until the reset is complete, so the driver uses it only if the target reset did not help.

MK