Date: Mon, 14 Jul 2003 21:45:36 -0600 From: Scott Long <scottl@freebsd.org> To: Ray Taft <ray.taft@jupiterhosting.com> Cc: freebsd-scsi@freebsd.org Subject: Re: Adaptec 2120S Errors Message-ID: <3F1378E0.8030401@freebsd.org> In-Reply-To: <GJEEKNFNPLOBKFKHOKGJOEKPCGAA.ray.taft@jupiterhosting.com> References: <GJEEKNFNPLOBKFKHOKGJOEKPCGAA.ray.taft@jupiterhosting.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Ray Taft wrote: > I have been beating my head on this for a few days, any help would be > appreciated. > > HARDWARE: > > Dual XEON 2.4Ghz - Hyperthreading = Off > 3GB RAM - ECC Registered > 3 X 15,000 RPM SEAGATE HARD DRIVES - Half Height > Adaptec 2120S Controller - Flashed to latest version > RAID 5 Configured - The system has completed the verify / build array > process (100%) weeks ago. > > All of the hardware is brand new (less than a month old), and we have sprung > for more expensive shielded SCSI cables. > > Here is the boot output on the controller initialization. Please note - this > 2120s does NOT have an optional battery (which the system says it does) and > has a 64MB cache where the system reports only 48MB. > The errant battery message is a known problem; I need to sit down some time and figure it out. It's only informational and is completely harmless. The 64MB vs 48MB disparity is due to the way the i960-based cards handle their memory. They reserve 16MB out of the cache for their own needs. So, the value reported reflects how much cache is usable, not the amount of RAM on the board. The StrongArm-based cards have RAM dedicated to the processor and make the full amount of cache available. > Jun 25 01:44:27 lease042 /kernel: aac0: <Adaptec SCSI RAID 2120S> mem > 0xf8000000-0xfbffffff irq 16 at device 1.0 on pci3 > Jun 25 01:44:27 lease042 /kernel: aac0: i960RX 100MHz, 48MB cache memory, > optional battery present > Jun 25 01:44:27 lease042 /kernel: aac0: Kernel 4.0-0, Build 6003, S/N b7d58a > Jun 25 01:44:27 lease042 /kernel: aac0: Supported > Options=1f7e<CLUSTERS,WCACHE,DATA64,HOSTTIME,RAID50,WINDOW4GB,SOFTERR,NORECO > ND,SGMAP64,ALARM,NONDASD> > [...] > > ATTMEPTED REMIDYS: > > We have swapped out the cable twice with different brands with different > shielding. No effect. We do not use a converter to change pins. Strait U320 > to U320 rated hardware and cables. Are there any hot-swap backplanes involved? Many backplanes out there perform marginally at u320. > > Rebuilt 4.8 kernel from scratch - twice. No effect. > This likely will have no effect. > Moved SCSI card to #1 PCI-X slot on motherboard. > This likely will have no effect except to slow down your PCI-X bus to PCI-64/66. > Researched 2120s history with FreeBSD as well as U320. > > Rebuilt and reverified RAID 5 array. Per a past support post, we tried > turning on and off write cache. For the errors below, write cache is turned > ON. You might want to investigate the on-drive caches also. It can be configured from the BIOS. > > Waited to install OS until the build / verify process completed for the RAID > 5 array off controller. > > ERRORS: > > We are seeing the following messages in /var/log - dmesg on a daily basis. I > understood this hardware had some buggy embedded code, but was supposed to > be fixed with the flash (six months ago). Also saw posts regarding this card > and its inability to operate under 4.7 and 4.8-RELEASE and was the reason we > put the box up on 4.8-STABLE. > > Here are the errors: As Don Lewis pointed out, your power-supply might be under powered. Unless the drives are in a backplane, it should be pretty easy to toss a second PS into the mix and test it out. Also, what make and model of drives are these? Have you checked the vendor for the latest firmware? Scott
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3F1378E0.8030401>