[dm-devel] qlogic failover

[gfbarros at ualr.edu]

And let me guess, hp strips the failover code out of their version of the firmware? :)

im having some trouble finding documentation on where to set the failover stuff up. we have a pair of the HP HSV100's as Patrick mentioned, so im hoping things will work as expected. Is there documentation on all this somewhere that im overlooking? otherwise, would either of you mind letting me know how to go about this?

thanks,
-guil


----- Original Message -----
From: Duane Grigsby <duane.grigsby at qlogic.com>
Date: Tuesday, June 15, 2004 1:37 pm
Subject: RE: [dm-devel] qlogic failover

> Actually, it works both ways. For a vendor+model match, we will 
> issue the notify specified in the devtbl (i.e. 
> FO_NOTIFY_TYPE_SPINUP) or if a global notify is specified we will 
> issue it for any device when a failover occurs, so it does allow 
> you to mix arrays as long as you don't specify the global notify 
> parameter. 
> 
> Duane
> 
> > -----Original Message-----
> > From: dm-devel-bounces at redhat.com 
> > [mailto:dm-devel-bounces at redhat.com]On
> > Behalf Of Patrick Mansfield
> > Sent: Tuesday, June 15, 2004 11:09 AM
> > To: device-mapper development
> > Cc: Andrew Vasquez
> > Subject: Re: [dm-devel] qlogic failover
> > 
> > 
> > On Mon, Jun 14, 2004 at 01:42:59PM -0500, Guilherme Barros wrote:
> > > is the qlogic failover setup in their driver aware of 
> > having to start 
> > > controllers showing ghost paths if the master one fails?
> > 
> > > let me try that one again. say controller A is showing actual 
> paths 
> > > (sda, sdc), and controller B is showing ghost paths (sdb, sdd). 
> say 
> > > controller A dies... does the qlogic driver with failover 
> > know that it 
> > > needs to activate controller B ?
> > 
> > Is these the COMPAQ HSV and MSA devices that need the spinup?
> > 
> > The qlogic driver conditionally sends a spinup for the
> > FO_NOTIFY_TYPE_SPINUP. AFAICT it is all or nothing since it 
> > is based on
> > the global variable qla_fo_params.FailoverNotifyType, so mixing 
> arrays> where some require spinup and some don't could break it.
> > 
> > # grep FO_NOTIFY_TYPE_SPINUP *.[hc]
> > qla_cfg.c:      } else if (qla_fo_params.FailoverNotifyType ==
> > FO_NOTIFY_TYPE_SPINUP ||
> > qla_cfg.c:                      old_lp->fcport->notify_type ==
> > FO_NOTIFY_TYPE_SPINUP ){
> > qla_devtbl.h:   {"COMPAQ", "MSA1000", 2, FO_NOTIFY_TYPE_SPINUP, 
> > qla_devtbl.h:   {"COMPAQ", "HSV110 (C)COMPAQ", 4, 
> > FO_NOTIFY_TYPE_SPINUP,   
> > qla_devtbl.h:   {"HP", "HSV100", 4, FO_NOTIFY_TYPE_SPINUP,   
> > qla_fo.c:       if (qla_fo_params.FailoverNotifyType ==
> > FO_NOTIFY_TYPE_SPINUP ||
> > qla_fo.c:               new_lp->fcport->notify_type ==
> > FO_NOTIFY_TYPE_SPINUP ) {
> > qlfo.h:#define FO_NOTIFY_TYPE_SPINUP                  5
> > qlfo.h:#define FO_NOTIFY_TYPE_DEF                
> FO_NOTIFY_TYPE_SPINUP> qlfolimits.h:#define FO_NOTIFY_TYPE_SPINUP   
>                 5
> > 
> > -- Patrick Mansfield
> > --
> > dm-devel mailing list
> > dm-devel at redhat.com
> > https://www.redhat.com/mailman/listinfo/dm-devel
> > 
> 
> --
> dm-devel mailing list
> dm-devel at redhat.com
> https://www.redhat.com/mailman/listinfo/dm-devel
>