Red Hat Bugzilla – Bug 37531
The installer cannot format and install RH 7.1 using AMI Megaraid 1600 (493) controller. Installs fine on IDE-drive
Last modified: 2007-04-18 12:32:50 EDT
From Bugzilla Helper: User-Agent: Mozilla/4.0 (compatible; MSIE 6.0b; Windows NT 5.0) I am trying to install RH 7.1 onto a server using the AMI MegaRAID 1600 (493) card. The card gets detected, the drive array gets detected. After selecting the relevant packages, I go into using DiskDruid. However, I get the message that I have to initialize /dev/sda first. For this message to disappear, OK has to be pressed five times before you enter DiskDriud. I then select my partitions, and finally I press OK in order to install. The installer returns the message "Could not format /. This may be a serious error" and I have to reboot. Installing to an IDE-drive works fine, installing to the RAID-array using RH 7.0 works fine. Reproducible: Always Steps to Reproduce: 1. Boot and start the installer 2. Select Server or Custom install 3. If custom, select packages and partitions (using fdisk og DiskDruid) 4. Press OK in order to install 5. You will get the message "Could not format /. This may be a serious error" Actual Results: I had to reboot the computer, leaving the RAID-array in the same state as before installing. Expected Results: Since RH 7.0 has a driver for the MegaRAID 1600 (493) controller, I assume that 7.1 also has this. The controller is detected and everything seems fine, but it just doesn't work. RH 7.0 works like a dream, but when I try to compile the 2.4.3-ac9 kernel on this one, I get the same error (sda: unknown partition table, VFS: Cannot open root device "801" or 08:01 etc...) The server is otherwise a standard box from Intel, dual P3-1ghz, 512mb RDIMM, AIC 7899, AMI Megaraid 1600 (493) with 5x36gb 10k drives.
I have exactly the same problem using an AMI MEGARAID 493 (Elite 1600) controleur with 64Mb onboard and RAID1 2x9Go. In kernel 2.4.1, 2.4.2 and 2.4.3, the MEGARAID driver same to be broken. All is perfect with Redhat 7.0 and 2.2 kernel.
Same problem with 128Mb onboard "RAID 5" 4x18Gb. 2.2.x --> OK 2.4.x --> sdb: unknown partition table
AMI help us to find what's wrong : > -----Message d'origine----- > De : Brian Highers [mailto:BrianH@ami.com] > Envoyi : vendredi 27 avril 2001 01:24 > @ : 'Loic.Breat@INTRINsec.com' > Cc : Peter Jarrett; Atul Mukker.; Venkatesh Ramamurthy > Objet : RE: megaraid driver doesn't work with 2.4.x kernels > > > Enclosed is a newer FW G158. The issue that you reported as > well as some > of the other posts is a known problem with FW L148. Please > upgrade to this > newer firmware and this should address your issues. > > Thanks > BrianH Using the new MEGARAID FIRMWARE G158 solve the problem. I can sent the firmware by mail if someone want it. Regards from France, Chris.
The last firmware version G158 is not available in the web http://www.megatrends.com/ (I can only find C158,-311 L148-311,H132-307) Where can I get the last version for my card Ami Elite 1600? How can I install the firmware? (There is no information about this neather) Do I need windows to do this? I hope to get information soon, this production machine needs to be in production as soon as possible and i can not use a 2.2.x kernel for this (Files > 2GB) Sincerely Rafael Martinez
I don't know when AMI will release a fixed firmware. However, the Red Hat Linux 2.2.19 kernel does support files > 2 Gb.
Standard kernel, or just the enterprise one? It used to be that only the enterprise kernel had LFS patches
I used C158 from AMI's site and it appears to have fixed it - at least enough to complete the install. Was G158 a typo?
No, i really have a G158. This is the release notes from G158 : AMI MegaRAID Firmware Image Release OEM Name: AMI Controllers Supported: 475/493/471/494/503/463/475LC Engineering Release date: 04/13/2001 Released by: Basavaraj / Chayan New Release Version: G158 BIOS Version: 3.11 Ctrl+M Version: 5.20 WebBIOS Version: 1.38 Boot ROM Version: 1.2.7 (AMI specific) Qlogic 12160 F/w version: 11.12.130 (11:32 Apr 02, 2001) Previous Official Release: F158/3.11/5.20/1.38 Controllers Supported: ====================== Series 463 = MegaRAID Express 550 Series 471 = MegaRAID Enterprise 1600 Series 475 = MegaRAID Express 500 Series 475LC = MegaRAID Express 500LC Series 493 = MegaRAID Elite 1600 Series 494 = MegaRAID Elite 1650 Series 503 = MegaRAID Enterprise 1700 Firmware Changes ================ Changes in G158 (based on F158 below) 1. QLogic firmware changed from 11.12.13 to 11.12.130 NEC471PR010 2. From Russel Foster: Here is the updated bbu493 code which has the period in between monitoring cycles extended. This should eliminate the recurrent messages. NEC471021 3. Recovery Sequence: When Medium Error occurs on Read, the RAID F/W recovers the data from other disks and fires a WRITE_VERIFY command. Due to a bug, it was issuing WRITE_VERIFY only if the channel is a fibre channel. It is now fixed to issue WRITE_VEIFIFY to scsi channels too. -----------------------------------------------------------------------
It seems that with the G158 firmware, we have many "READ ERRORS" problem on the LOGICAL Drive :((((((( We will try the C158 firmware soon. bouuuuu....
All appears to be ok with FIRMWARE C158. no more I/O Errors. Chris.
AMI recommends everybody who uses the "148" firmware to immediately upgrade to a 158 version, the problems are indeed purely firmware based.
Further confirmation... Yes, I had the same problem trying to install RH 7.1 onto my system which is using the Elite 1600 RAID card as its sole disk controller. Installation would complain that it couldn't figure out the "unknown partition table" and exited. Following advice found here, I downloaded the C158 firmware upgrade from the download page at http://www.megaraid.com, performed the upgrade (which was a little tricky loading on a floppy along with the system files required for booting -- I dumped the readme.txt file and "drivespace.sys" [I think it was or some similarly named thing]), and the 7.1 release installed just fine from the CDs. -- Fred P.
I'd like to close this as "NOTOURBUG", but that doesn't exist -> "NOTABUG"