I think I've found the answer. After briefly testing the new card with an
external dual-drive enclosure, I simply removed the drives from the bus
without rebooting. So, the device table still said there were two disks
there -- thus the message was issued twice. I just now reattached the
drives and powerdown worked just fine. Removed them, and got the same
error messages. Issuing /sbin/scu scan edt seems to be the fix.
The question follows:
We recently installed a KBPZA-CA PCI to UltraSCSI host adapter card into
a new XP1000 running DU4.0E. It was briefly tested by attaching an
external drive, mounting a disk, and writing/deleting a small file. There
is currently nothing attached to the card. There are two internal drives
on the internal scsi bus: root and an unused disk. The kernel was rebuilt
to recognize the new card.
After rebooting, /var/adm/messages reports
May 11 17:14:35 alpha10 vmunix: cdisk_stop_unit_done: [1/1/0] Max spindown
attempts made.
May 11 17:14:35 alpha10 vmunix: Device will no longer be power managed.
May 11 17:14:35 alpha10 vmunix: cam_disk: [1/1/0] Failed to deregister
from power manager (0x1)
May 11 17:15:05 alpha10 vmunix: cdisk_stop_unit_done: [1/2/0] Max spindown
attempts made.
May 11 17:15:05 alpha10 vmunix: Device will no longer be power managed.
May 11 17:15:05 alpha10 vmunix: cam_disk: [1/2/0] Failed to deregister
from power manager (0x1)
root has these entries in its crontab file:
0 18 * * 1-5 /sbin/sysconfig -r pwrmgr disk_spindown=1 > /dev/null
30 8 * * 1-5 /sbin/sysconfig -r pwrmgr disk_spindown=0 > /dev/null
and we see the above error messages repeated every day at about 18:20 (20
minutes is the idle time spindown setting).
A little testing with dxpower to enable disk spindown after 1 minute
begets the same messages in the dxconsole window, and the expected
10-15 second lag in response time for the disks doesn't happen.
Spindown worked before installing the new card (or, at least, there are
no such error reports in /var/adm/messages), and ddr_config verifies that
the disks' PwrMgmt_Capable = true.
Does anyone have any ideas on what's going on, and how to restore the
spindown capability? Could this new card be interfering with the internal
bus? I don't find anything about power management in the card's EEROM
parameter settings tables.
Received on Thu May 20 1999 - 15:38:47 NZST