sa0 devices disappear after power cycling tape library

[Mark Saad]

On Mar 14, 2014, at 10:15 PM, Dan Langille <dan at langille.org> wrote:

>>> On Mar 14, 2014, at 8:14 PM, Dan Langille <dan at langille.org> wrote:
>>> =20
>>> I have a tape library which acts up from time to time and I power cycle it=
>> .
>>> =20
>>> After running camcontrol devlist rescan, some of the devices are missing
>>> =20
>>> This what I see in dev after power cycling the FreeBSD 9.2 server:
>>> =20
>>> $ ls *sa*
>>> lrwxr-xr-x  1 root  wheel        6 Mar  8 02:12 esa0 -> esa0.0
>>> crw-rw----  1 root  operator  0x70 Mar  8 02:12 esa0.0
>>> crw-rw----  1 root  operator  0x76 Mar  8 02:12 esa0.1
>>> crw-rw----  1 root  operator  0x79 Mar  8 02:12 esa0.2
>>> crw-rw----  1 root  operator  0x7c Mar  8 02:12 esa0.3
>>> crw-rw----  1 root  operator  0x75 Mar  8 02:12 nsa0.1
>>> crw-rw----  1 root  operator  0x78 Mar  8 02:12 nsa0.2
>>> crw-rw----  1 root  operator  0x7b Mar  8 02:12 nsa0.3
>>> crw-rw----  1 root  operator  0x74 Mar  8 02:12 sa0.1
>>> crw-rw----  1 root  operator  0x77 Mar  8 02:12 sa0.2
>>> crw-rw----  1 root  operator  0x7a Mar  8 02:12 sa0.3
>>> =20
>>> This is what I see after rebooting the FreeBSD server:
>>> =20
>>> $ ls -l *sa*
>>> lrwxr-xr-x  1 root  wheel        6 Mar 15 00:09 esa0 -> esa0.0
>>> crw-rw----  1 root  operator  0x70 Mar 15 00:09 esa0.0
>>> crw-rw----  1 root  operator  0x76 Mar 15 00:09 esa0.1
>>> crw-rw----  1 root  operator  0x79 Mar 15 00:09 esa0.2
>>> crw-rw----  1 root  operator  0x7c Mar 15 00:09 esa0.3
>>> lrwxr-xr-x  1 root  wheel        6 Mar 15 00:09 nsa0 -> nsa0.0
>>> crw-rw----  1 root  operator  0x6f Mar 15 00:09 nsa0.0
>>> crw-rw----  1 root  operator  0x75 Mar 15 00:09 nsa0.1
>>> crw-rw----  1 root  operator  0x78 Mar 15 00:09 nsa0.2
>>> crw-rw----  1 root  operator  0x7b Mar 15 00:09 nsa0.3
>>> lrwxr-xr-x  1 root  wheel        5 Mar 15 00:09 sa0 -> sa0.0
>>> crw-rw----  1 root  operator  0x6e Mar 15 00:09 sa0.0
>>> crw-rw----  1 root  operator  0x74 Mar 15 00:09 sa0.1
>>> crw-rw----  1 root  operator  0x77 Mar 15 00:09 sa0.2
>>> crw-rw----  1 root  operator  0x7a Mar 15 00:09 sa0.3
>>> crw-rw----  1 root  operator  0x6d Mar 15 00:09 sa0.ctl
>>> =20
>>> Should I just do some ln -s?  Something easier, I=E2=80=99m sure=E2=80=A6.=
>> 
>>> =20
>> 
>> 
>> Have you tried a camcontrol rescan ?
> 
> Yes, that’s what I meant by 'camcontrol devlist rescan’ above…
> 

Sorry I missed that , I think that syntax may not do what you want 

> I ran: camcontrol rescan all.  The output above is after that command.
> 


How is the dlt drive attached , fc-al , sas or SCSI ? 


>> Also what does camcontrol devlist -v   s=
>> how before and after the reboot ?=20
> 
> I don’t have before, but after is:
> 
> $ sudo camcontrol devlistol devlist
> Password:
> <ATA TOSHIBA DT01ACA3 ABB0>        at scbus0 target 2 lun 0 (pass0,da0)
> <ATA TOSHIBA DT01ACA3 ABB0>        at scbus0 target 10 lun 0 (pass1,da1)
> <ATA TOSHIBA DT01ACA3 ABB0>        at scbus0 target 11 lun 0 (pass2,da2)
> <ATA TOSHIBA DT01ACA3 ABB0>        at scbus0 target 12 lun 0 (pass3,da3)
> <ATA TOSHIBA DT01ACA3 ABB0>        at scbus0 target 13 lun 0 (pass4,da4)
> <ATA TOSHIBA DT01ACA3 ABB0>        at scbus0 target 14 lun 0 (pass5,da5)
> <ATA TOSHIBA DT01ACA3 ABB0>        at scbus0 target 16 lun 0 (pass6,da6)
> <ATA TOSHIBA DT01ACA3 ABB0>        at scbus0 target 17 lun 0 (pass7,da7)
> <TOSHIBA DT01ACA300 MX6OABB0>      at scbus5 target 0 lun 0 (ada0,pass8)
> <TOSHIBA DT01ACA300 MX6OABB0>      at scbus5 target 1 lun 0 (ada1,pass9)
> <QUANTUM DLT8000 0250>             at scbus7 target 4 lun 0 (sa0,pass10)
> <OVERLAND LXB 0524>                at scbus7 target 6 lun 0 (pass11,ch0)
> 
> IIRC, before I ran the command, the last two entries were missing.
> 

So this is a robotic library and drive ? 


Also do I I read you right the sa0 entry is back,or you still just have have sa0.0 ?

  You could use etc/devfs.conf to make a link entry , see the end of the file for examples. But I suspect something else is up . I have seen this before but I can't remember what caused it off hand . 


> -- 
> Dan Langille - http://langille.org
> 


Mark saad | mark.saad at longcount.org