Date: Sat, 15 Feb 2014 14:14:29 -0700 From: Scott Long <scott4long@yahoo.com> To: Karl Pielorz <kpielorz_lst@tdx.co.uk> Cc: "FreeBSD-scsi@freebsd.org" <freebsd-scsi@freebsd.org> Subject: Re: Wiring down devices with LSI 2308 / mps? Message-ID: <BEE9554F-182B-4FDB-B86E-12AB766749F6@yahoo.com> In-Reply-To: <41337B9333D94F88B9469A60@study64.tdx.co.uk> References: <41337B9333D94F88B9469A60@study64.tdx.co.uk>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi, The LSI cards keep a persistent mapping of hard drive serial numbers to = target numbers. That way, when it=92s in RAID mode and you pull a drive = from one slot and put it into another slot, it=92ll recognize that it=92s = the same drive and part of the same array set. It=92s incredibly = annoying, and once the persistent mapping is created, it cannot be = cleared just by turning off the RAID mode. What we do at Netflix is use = the a tool provided by LSI to clear out the persistent mapping (and all = of NVRAM, actually), and then turn off the persistence feature. = Unfortunately, you=92ll need to contact LSI directly for this tool. = Once you have it, I can share a script that performs the necessary = actions. Contact me off-list if you=92d like more information. Scott On Feb 15, 2014, at 9:14 AM, Karl Pielorz <kpielorz_lst@tdx.co.uk> = wrote: >=20 > Hi, >=20 > I've got a SuperMicro board that has an onboard LSI 2308 onboard. This = appears using the mps driver: >=20 > " > mps0: <LSI SAS2308> port 0xc200-0xc2ff mem = 0xf33c0000-0xf33cffff,0xf3380000-0xf33bffff irq 44 at device 7.0 on pci0 > mps0: Firmware: 15.00.00.00, Driver: 16.00.00.00-fbsd > mps0: = IOCCapabilities:185c<ScsiTaskFull,DiagTrace,SnapBuf,EEDP,TransRetry,IR> > " >=20 > I'd like to wire the devices down on this - but that doesn't seem to = work :( >=20 > I tried this by putting entries in '/boot/device.hints' such as: >=20 > " > hint.scbus.0.at=3D"mps0" >=20 > hint.da.0.at=3D"scbus0" > hint.da.0.target=3D"2" > hint.da.0.unit=3D"0" >=20 > hint.da.1.at=3D"scbus0" > hint.da.1.target=3D"1" > hint.da.1.unit=3D"0" > ... > " >=20 > Testing though shows that maybe the controller itself is collapsing = 'missing' ports/targets? >=20 > e.g. Attach a disk to 'Port 0' - and it shows as: >=20 > da2 at mps0 bus 0 scbus0 target 2 lun 0 >=20 > Ok, target 2? - but at least that's mapped to da2 (wiring works) >=20 > Take the disk off, and connect it to Port 1 - and it shows as: >=20 > da2 at mps0 bus 0 scbus0 target 2 lun 0 >=20 > Port 1 is target 2 as well? - Switch the drive to 'Port 2' - and you = get: >=20 > da2 at mps0 bus 0 scbus0 target 2 lun 0 >=20 > So port 2 is also target 2? (if the others aren't occupied) >=20 > Add a 2nd drive (to any port) and you get: >=20 > da3 at mps0 bus 0 scbus0 target 3 lun 0 >=20 >=20 > So the 'wiring down' is working - but it looks like the LSI collapses = it's own 'missing drives'? - And starts at Target 2? >=20 > I can't see any setting in the LSI bios that would change this? - I'm = presuming it must be possible not to have it do this - though searching = the lists didn't turn anything obvious up... >=20 > Thanks, >=20 > -Karl >=20 > _______________________________________________ > freebsd-scsi@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-scsi > To unsubscribe, send any mail to = "freebsd-scsi-unsubscribe@freebsd.org"
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?BEE9554F-182B-4FDB-B86E-12AB766749F6>