controller errors from RELENG_9 to RELENG_10 upgrade

[Mike Tancsa]

On 8/30/2014 12:24 AM, Glen Barber wrote:
>
> I am suspicious of the CAM changes between stable/9 and stable/10.  Did
> you rebuild all ports on this system after the upgrade?  smartctl(8)
> reflects 10.1-PRERELEASE, so I am particularly curious of anything that
> uses hald(8) or dbus(8).

Hi Glen,
	I have rebuilt.  I am wondering if this is really a hardware issue 
somewhere. In the mean time, I had deleted a number of old files and 
snapshots that reclaimed space and the errors have greatly diminished. 
Is it possible these errors are made worse when one drive in a PMP slows 
things down with a bunch of write errors.

On another note, the other issue I have noticed is at boot up time, the 
last set of disks on pmp2 does not scan until late in the boot process.

What is the best way to work around this, so all is scanned before zfs 
tries to mount things. In dmesg, I see a

(noperiph:siisch1:0:-1:ffffffff): rescan already queued
(noperiph:siisch3:0:-1:ffffffff): rescan already queued


so later on I see ....

Trying to mount root from ufs:/dev/da2s1a [rw]...
ZFS filesystem version: 5
ZFS storage pool version: features support (5000)
pmp2 at siisch0 bus 0 scbus0 target 15 lun 0
pmp2: <Port Multiplier 47261095 1f06> ATA-0 device
pmp2: 300.000MB/s transfers (SATA 2.x, NONE, PIO 8192bytes)
pmp2: 5 fan-out ports
ada12 at siisch0 bus 0 scbus0 target 1 lun 0

I had to do a /etc/rc.d/zfs restart once the box was up.



-- 
-------------------
Mike Tancsa, tel +1 519 651 3400
Sentex Communications, mike at sentex.net
Providing Internet services since 1994 www.sentex.net
Cambridge, Ontario Canada   http://www.tancsa.com/