The SCSI layer should retries a failed I/O 5 times, and have 30 second timeouts per I/O. The card itself my return the I/O as failed before the 30 second timeout, which would improve your failover times. That gives 3 minutes as a maximum. Thanks -steve On Fri, 2003-12-12 at 16:35, Franc Carter wrote: > Thanks, > > can you give me a ballpark on 'high' >5 minutes ? > > cheers > > On Sat, 13 Dec 2003 04:35 am, Steven Dake wrote: > > The multipath code is simple and seems pretty stable to me. > > > > There is one issue, in that failover can sometimes take a long time > > because of the SCSI layer's insistence upon executing error recovery. > > In the case of multipath I/O's, there is no need to error recover before > > processing the I/O. Instead, once error is detected, failover, then > > start error recovery. > > > > The SCSI layer doesn't do this currently, so failover times are high. > > > > Thanks > > -steve > > > > > > On Thu, 2003-12-11 at 21:52, Franc Carter wrote: > > > Hi, > > > > > > I've been using linux sofwtare raid for years and had no problems. > > > We are about to install some disk systems that have dual contollers > > > and am wondering wether multipath is as stable as the rest of software > > > raid ? > > > > > > thanks > > > > - > > To unsubscribe from this list: send the line "unsubscribe linux-raid" in > > the body of a message to majordomo@vger.kernel.org > > More majordomo info at http://vger.kernel.org/majordomo-info.html > > > > - To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
