Re: send scsi commands Just for the records and benefit of others:
The problem was/is the 5712 adapter who has big problems talking to some
SCSI devices.
The supplier of the archiving software has pulled support for this adapter
and is actively working with IBM to resolve the issues.
Meanwhile we use a 6203 and it works like a charm.
Thanks for all the input, I leared quite a bit.
/gerold
"Gerold Krommer" <g.krommer@doremove.fns.co.at> schrieb im Newsbeitrag
news:1126099972.699796@newsmaster-03.atnet.at...
> Hi all,
>
> I have been scratching my head on this for a week now (hair is gone) and
> now I'm turning to this community.
>
> We sell (not engineer) an archiving software that talks to optical
> jukeboxes over scsi.
> The platform in question is a p5 LPAR, AIX 5.2 ML6, 5712 dual channel SCSI
> Adapter, HP 1900ux LVD SCSI jukebox
>
> When looking at the system the 4 odm devices are created.
>
> A configuration utility of ours (that installes special drivers) does NOT
> see the devices. What this utility does is scan through all SCSI busses
> sending a TUR (test unit ready). Interestingly this returns errno 19 (No
> such device) for SCSI Ids that have no devices attached, which is correct,
>
> This utility returns errno 5 (IO error) for those SCSI Ids that have
> devices attached (4drives and 1 robotic arm).
>
> To make a long story short I'm looking for strategies to narrow down the
> faulty component. I have a few ideas for the HW, but would also like to
> eliminate the config utility as a source of problem.
>
> I'm looking for a program that allows me to send SCSI commands to a
> specific address of a SCSI bus. I browsed/googled for hours, but it seems
> only vendors of backup software (like Legato or NSR) provide such tools
> with their software (like scsi_command)
>
> Does anyone have such a utility or code they would like to share? I looked
> and maybe it will take me a few days to wade through the SCSI structures
> and use ioctl to do it myself, but why re-invent the wheel?
>
> Thanks for any help for the desparate,
>
> /Gerold
> |