Greetings. We have DU4.0D and DU4.0E systems (GS140s and 4100s) that
are generating continuous errors since we made some mods on our
ESA10000 array with HSZ70 controllers. The error logs talk about
devices with references to bus #, target #, and lun #, but we
are having a difficult time translating these numbers into actual
devices.
Our configuration is a group of 4 systems, all connected thru a
5-port hub to the bottom half of the ESA10000, which is standalone,
not chained to the top half. The chart below shows bus/target/luns
erroring when all systems were attached to the ESA10000, and
bus/target/luns still complaining after disconnecting the "tri-link"
(end of cable with terminator, attached to hsz70) from the
controllers.
Original errors Errors after removing tri-link
sys1: bus 1, target 7, lun 7 bus 1, target 1, lun 2
bus 1, target 0, lun 0
bus 1, target 1, lun 2
sys2: bus 1, target 7, lun 7 bus 2, target 4, lun 0
bus 1, target 1, lun 2
bus 2, target 4, lun 0
sys3: bus 5, target 7, lun 7
bus 5, target 1, lun 2
sys4: bus 3, target 7, lun 7
My field service person can't make heads or tails of the */7/7 numbers,
and isn't really sure what the other ones are either. Can anyone
tell us how to translate these to real devices, given that each system
has 2 regular scsi controllers, and 2 scsi controllers connected to
hsz70s, and on the gs140s the controllers are spread between two PCI
enclosures, with separate hoses to each PCI enclosure???
We are suspicious that "sys1" is generating controller errors which
are causing scsi bus resets that are upsetting all other systems.
"sys1", however, has its controller set to disable resets, so it's
not that simple. Before pulling the tri-link, "sys1" was generating
10 errors a minute, the others about 1 a minute, with it disconnected
"sys1" is still throwing 1 error a minute.
Thanks for whatever insights you can send!
--
Judith Reed
jreed_at_appliedtheory.com
(315) 453-2912 x335
Received on Wed Oct 20 1999 - 21:06:33 NZDT