Hello mlaspina ,

A network trace was already attached, see my second message above.

In the meantime we found the "real" cause thanks to Michael (etherboot): 

He checked the trace and told me:

iSCSI targets have a habit of  sending a spurious CHECK CONDITION response to 
the first SCSI command, with  sense data indicating that "power-on occurred".  
We already issue and ignore 
one extra READ CAPACITY (10) command to draw out this response.

>From the capture that you provided, it seems that the comstar target is 
>sending not one but two spurious CHECK CONDITION responses: one saying 
>"power-on occurred" and one saying "reported LUNS data has changed".   We 
>currently expect only one spurious response, so the second spurious response 
>gets treated as a genuine error by gPXE.

Now Michael updated GPXE to attempt up to 10 dummy commands at start of day 
before assuming that error responses are genuine.

Today I tested latest GPXE and confirmed Michael was correct.

kristof
-- 
This message posted from opensolaris.org
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss

Reply via email to