Re: Ignoring devices

[mlelstv%serpens.de@localhost (Michael van Elst)]

kab00m%lich.phys.spbu.ru@localhost (Dima Veselov) writes:

>08.01.2020 10:44, Martin Husemann пиÑ?еÑ?:
>> On Wed, Jan 08, 2020 at 01:56:41AM +0300, Dima Veselov wrote:
>>> are not accessible to the box. According to standard FC
>>> behaviour HBA have all devices enumerated on the bus,
>>> but can not read them.
>> 
>> Can you show example dmesg output for such a device?

>sd0 is primary working device on array controller A, sd1 is the device 
>on controller B,
>which is visible, but not readable. Disk array require special license 
>to map different
>LUNs to different hosts and I don't have one, therefore all LUNs are 
>mapped to all
>hosts. That cause an issue.

Yes, when one host reserves a unit, the other gets a failure.


>Jun 10 16:08:17 ranir /netbsd: sd1(isp0:0:0:1):  Check Condition on CDB: 
>0x08 00 00 00 01 00
>Jun 10 16:08:17 ranir /netbsd:    SENSE KEY:  Illegal Request
>Jun 10 16:08:17 ranir /netbsd:     ASC/ASCQ:  ASC 0x94 ASCQ 0x01
>Jun 10 16:08:17 ranir /netbsd: sd1d: error reading fsbn 0 (sd1 bn 0; cn 
>0 tn 0 sn 0)

The disks are not shared, so this _should_ only be an issue when
both hosts boot at the same time. The only way to avoid this would
be a kernel that has all SCSI disks enumerated in its configuration,
which is not very practical.

The daily errors are caused by the /etc/security script (called by
/etc/daily).  All disks are accessed by the /etc/security script to
keep backups of disklabels and partition tables. You could either
disable that backup function (check_disklabels=no) or modify the
script to select only particular disks. The line is:

        disks="$(/sbin/sysctl -n hw.iostatnames)"

Another method is to detach all unused disks when booting, e.g.
by calling something like 'drvctl -d sd1' in /etc/rc.local.


-- 
-- 
                                Michael van Elst
Internet: mlelstv%serpens.de@localhost
                                "A potential Snark may lurk in every tree."