Date: Thu, 15 Nov 2007 16:22:28 -0500 From: "Simon" <simon@optinet.com> To: "Brian A Seklecki (Mobile)" <bseklecki@collaborativefusion.com> Cc: Sean McAfee <smcafee@collaborativefusion.com>, Jason Thomson <jason.thomson@mintel.com>, Benjie Chen <benjie@addgene.org>, "freebsd-hardware@freebsd.org" <freebsd-hardware@freebsd.org> Subject: Re: PERC5 (LSI MegaSAS) Patrol Read crashes Message-ID: <20071115212026.5C3E613C45A@mx1.freebsd.org> In-Reply-To: <1195160114.4042.154.camel@new-host>
next in thread | previous in thread | raw e-mail | index | archive | help
I don't find this as bashing, rather as valuable input, which I always welcome and appreciate! If nobody brings out the problems then how is one to know other than thru hard hands-on experience. Problem I see there is nothing better which works with FreeBSD when it comes to new servers. Even older ASR driver is not supported under amd64 Older LSI MegaRAIDs are same crap with half functionality malfunctioning or not supported. So what works better than Dell PERC5+DARC5+Broadcom NICs? I have been using Supermicro + LSI controllers and that is also a bunch of hack. HP servers any better? If Patrol Reads are marketing bullshit, what do you use? consistency checking? How do you insure integrity of your RAID5 arrays without either? should parity get screwed up or drives develop bad blocks and a drive fails, the array will either not have enough parity to rebuilt replaced drive or the entire array will get hosed. Happened at least few times to me during the last few years. The way I see it, until manufacturers such as Dell and HP start fully supporting FreeBSD, the mentioned problems will never go away and FreeBSD will always be a bunch of hack when it comes to supporting hardware such as RAID and or NIC cards. This is why partly many companies prefer Linux or even Windows. -Simon On Thu, 15 Nov 2007 15:55:14 -0500, Brian A Seklecki (Mobile) wrote: >On Wed, 2007-11-14 at 07:23 -0500, Simon wrote: >> Do you guys perform consistency checks on your RAID5 >> or other redundant arrays? >> >> -Simon >> >> >> On Wed, 14 Nov 2007 10:09:55 +0000, Jason Thomson wrote: >Normally I'd be praising Dell, but I think a little vendor bashing is >due here. >Its a software bug (driver). It can probably be easily fixed. I think >there's a PR on it somewhere (will check). >We eliminated firmware bugs through P.O.L. I suspect, but haven't >confirmed, that the driver developer used a reference platform w/ the >"Patrol Reads" disabled. Maybe that platform was provided by the >employer, having already passed through the hands of wise administrators >who saw through the marketing _BULLSHIT_ that "Patrol Reads" really are. >The linux beast-crew on poweredge-linux@ claims that it doesn't happen >on RHEL. But wow would they really be able to differentiate a >PERC5-related crash from any other inexplicable Linux kernel panic? >Well, there's a lot of them and they like to whine a lot, so I imagine >Dell would have done something. >The whole thing is a UGLY UGLY hack. Needing Linux 32-bit compat >support just to monitor and administer your RAID controller? Might as >well just install Redhat and send your root password to >misc@openbsd.org. >All of that functionality really should be tied into an out-of-band >system like the LOM. >The problem is that the DRAC5 is crap too, crap customized for Windows >users. >The PERC5 is OEM crappy closed source hardware, and Dell should know >better. >The Broadcom NICS (bge(4)) are crap. Let's not forget the HA NIC >features that only work in select OSs. >The only praise Dell gets is that the IPMI on the integrated IPMI/DRAC5 >card is faster. >It's all subtle vendor lock-in crap. As if no one remembers the 90s and >the fall of UNIX? >Dell selectively uses aspects of F/OSS methodology to suit their needs. >If you think back, 18 months after the 8th gen was release, FreeBSD was >rock-solid on that platform. Intel NICs and PERC4. AMI/LSI even >released the source code to the MegArc CLI. >Anyway, Dell really screwed up on the 9th gen. >~BAS (Likes to do his vendor bashing _before_ lunch, but sometimes takes >a late one)
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20071115212026.5C3E613C45A>