SCSI Errors filling messages FAST

[brian_31]

I have the following streaming errors going in my messages..kernel: SCSI error : <1 0 0 4> return code = 0x10000
Feb 7 9:22:37 vs80 kernel: end_request: I/O error, dev sdg, sector 0..the messages follows on. I think the 4 from 1 0 0 4 is the lun and it follows for 5 6 7 etc..like 1 0 0 5..looks like a SAN issue, could someone help?

Thanks

Brian

[Matti_Kurkela · Accepted solution]

If you have multipathing and all disks connected to a particular HBA are producing errors, you've probably lost a FibreChannel link. Perhaps a fiberoptic cable has been damaged, or a connector has come loose?

Knowing the name and version of your Linux distribution would be helpful. But here's some generic advice...

The first number in <1 0 0 4> etc. is the controller number. In your case, the HBA producing the errors is controller 1. If your system has a 2.6.* series kernel, run "ls -l /sys/class/scsi_host/host1" and see the target of the "device" symlink: you'll see the PCI ID of the HBA that is producing the errors. You can then use the "lspci" command to know the name of the HBA.

With 2.4.* or older kernels, run "less /proc/scsi/scsi" to get much of the same information in a different way.

After that, it's basic troubleshooting: check the link LEDs on the HBA(s), check the link status on FibreChannel switch(es) etc.

MK