Re: [foxboro] Software disk mirroring

On Tue, Feb 11, 2003 at 01:39:46PM -0500, William C Ricker wrote:
> 
> For one of our clients, Foxboro installed it on a 51E. This was done 
> back at version 6.1.

Thanks, I had a converation with someone at TAV this mornign that was
trying to tell me Foxboro no longer had a licsenese for this.

> 
> Since then, I don't believe any particular attention has been
> paid to it.  Based on recent work we have done for three other
> clients, I am convinced that it probably doesn't continue to
> work there, as it didn't with these other clients. These other 
> client's systems included AW51A, and AW51B, with 4.3.x, 6.2 and 
> 6.3 software.
> 
> Of the three, two had abandoned it; disconnecting the second
> drive and putting it on the shelf.  The third simply assumed
> it was running (wrongly).

Yes we originally added it's status to an operaator graphic, but that did
not work all that well, so we implemented email to our process control
group on failure. That has worked extremly well.

> 
> The problems start were dosumentation states errors are reported
> to the system monitor.  They aren't.  Problems are reported to
> the Sun/Solaris monitor, ending up with messages in the 
> 'messages' files under /var/adm, but most people never look there.
> 
> Different cards were used in different Sun boxes for the 
> 2nd SCSI interface.  Some software uses device addressing from 
> the older card while other uses addressing for the new one.
> 
> The command set, while simple, can be botched.  An error in command
> order can yield a 'disturbed' mddb data base replica set, causing
> both disks in the pair to be un-bootable, a fact that will only
> become known the next time you decide to reboot.  Some obvious
> errors in entering commands are not checked and reported or 
> confirmed. 
> 
> It is possible that the EE PROM address used in booting from the
> second drive (on failure of the first) may be incorrect.
> 
> Both the old (dump_8MM) and new (backup) dump-to-tape programs 
> have problems. The old one may look at the status of the pair 
> using the wrong device address.  The new one always gets it's 
> mddb slice from the primary drive.  If it's not there, a one 
> line error message in the midst of many lines of valid progess 
> messages easily gets lost, and the tape can't be used to produce 
> a bootable disk.  The documentation doesn't tell you you must
> only make tapes from the primary.
> 
> For yet another client, we have put together a display, script
> and compound to monitor the mirrored pair, and report/alarm
> errors.  This provides a base level of knowledge of when the
> function is actually working.
> 
> We can't be too calalier about modifying the standard Foxboro
> scripts as we (an integrator) are not in control of the update
> process for most of our customers.  We must generally impliment
> changes outside the standard modules, or by documentation or
> proceedure.  Hence, we have not looked at the script set for this
> function with any eye to correcting some of these problems.

As an end user, I sometimes feel that I'm not in control of this same
process :-). I don't have this feeling with other vendors :-(
> 
> And finally, there was a note about making a loaded, on the shelf
> backup disk by using the 'unmirror_sys' command.  Our testing 
> did not yield the backup reliably.  To be sure,  we could boot
> from the backup more often that not, but probably 3 or 4 out of 
> 10 tries would give us a disk that would not boot.  We tried this
> on AW51A at 4.3 and AW51B at 6.3.
> 
> Oh.  Wait a minute.  the original question was:
> 
> >> Has anyone been "allowed" by Foxboro to deploy Medusa based disk 
> >> mirroring on a 64 bit Sun box (D or E)?
> 
> The answer is , yes.  And they do install it for clients, as well.
> 

Thanks for the thoughtful reply.

Having had the "new improved" hardware raid solution take machines down on
an average of oce a week for the last couple of months, I'm wanting to go
back to this!


-- 
"They that would give up essential liberty for temporary safety deserve
neither liberty nor safety."
                                                -- Benjamin Franklin
 
 
_______________________________________________________________________
This mailing list is neither sponsored nor endorsed by Invensys Process
Systems (formerly The Foxboro Company). Use the info you obtain here at
your own risks. Read http://www.thecassandraproject.org/disclaimer.html
 
foxboro mailing list:             http://www.freelists.org/list/foxboro
to subscribe:         mailto:foxboro-request@xxxxxxxxxxxxx?subject=join
to unsubscribe:      mailto:foxboro-request@xxxxxxxxxxxxx?subject=leave

Other related posts: