[Info-vax] SCSI issues on Alpha

[Rich Jordan]

On Monday, December 23, 2013 10:09:47 AM UTC-6, Scott Dorsey wrote:
> I've got a user who is a great fan of his  Alpha Personal Workstation 433a,
> 
> and I'd like to consolidate the disks on it and replace them with a newer
> 
> mirrored pair of disks.
> 
> 
> 
> Currently on the SCSI chain he has the internal RZ1CC-BA, and externally
> 
> he's got one Seagate ST1510N drive and two IBM DNES-318350W drive.  So the
> 
> mix of wide and narrow is a little weird.
> 
> 
> 
> Now, the only newer SCSI drives I have are Seagate ST3146807LC, since new
> 
> scsi drives are not a common thing.  I can put them either on an adaptor
> 
> to an 80-pin Honda connector or an adaptor to a 50-pin Centronics-style
> 
> connector, but in either case the buss becomes extremely flakey... when 
> 
> I am at the monitor rom and do a "sh dev d" everything disappears or 
> 
> large numbers of duplicate drives appear.
> 
> 
> 
> So... my first question is why is the "sh dev d" on this system so different
> 
> than on the Decstation machines?  It seems like it is caching something; it
> 
> does not do a buss reset every time I run the command.  What do I need to do
> 
> in order to get it to actually force a buss interrogation?
> 
> 
> 
> And the second question is... why is this happening?  Is there something 
> 
> about the ST3146807LC that is a problem?  The low voltage mode?  The 
> 
> differential mode?
> 
> 
> 
> I tried moving the whole external chain over to an Ultrasparc machine and 
> 
> everything works just fine there, I can't make the buss do anything odd.
> 
> So I am thinking it's something specific to the way the Alpha implements
> 
> it.
> 
> 
> 
> Is there some other large drive I should be trying here?  Am I missing
> 
> something important?
> 
> --scott
> 
> -- 
> 
> "C'est un Nagra. C'est suisse, et tres, tres precis."

There are a lot of the 80-pin adapters (that allow connecting an SCA drive to a 50 or 68 pin adapter) that are not reliable, and won't work with many drives, especially LVD models.  I no longer have the info because I destroyed all my bad ones.  Basically I won't trust one with a 50 pin connector with any LVD drive (which the ST3146807LC is).

The PWS had a built in UltraSCSI HBA, which is not LVD.  LVD drives when connected would downshift into SE mode.  What kind of HBA was in the UltraSparc system?  If it was an LVD then you're comparing totally different environments.

Also is the HBA in the UltraSparc at a different ID than the PWS HBA?



I like the IBM 36 and 73GB 2.5" drives (ST973401LC for the 73GB).  They are 80-pin, and you can fit them into the cheap adapter trays that allow laptop drives to be used in PC desktops, and use a quality 80-pin to 68 pin adapter.  They're fast (for their age), low power, and pretty quiet.  And cheap on Ebay.  I've run them in an Alphastation 200 with a KZPBA controller (Ultra-Wide, similar to the controller in the PWS) as well as in separate enclosures with a KZPEA (Ultra-160) and got the expected levels of performance; I doubt you can use the KZPEA in the PWS but you could get acheap KZPBA Ultra-40 HBA in a spare slot and run your chain of faster drives off of it, leaving the embedded one to run the CDROM (if its SCSI) and any other narrow devices.