From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; Galeon) Gecko/20031202 Galeon/1.3.11a Description of problem: The default ips driver version is 6.00.00. The needed ips driver version is 6.11.xx. The default ips driver causes the following kernel-errors: Jan 6 14:32:14 dev-srv5 kernel: Vendor: IBM Model: SERVERAID Rev: 1.00 Jan 6 14:32:14 dev-srv5 kernel: Type: Direct-Access ANSI SCSI revision: 02 Jan 6 14:32:20 dev-srv5 kernel: (ips0) Reset Request - Flushed Cache Jan 6 14:32:40 dev-srv5 kernel: scsi: device set offline - not ready or command retry failed after host reset: host 1 channel 0 id 4 lun 0 Jan 6 14:32:40 dev-srv5 kernel: Vendor: IBM Model: SERVERAID Rev: 1.00 Jan 6 14:32:40 dev-srv5 kernel: Type: Direct-Access ANSI SCSI revision: 02 Jan 6 14:32:46 dev-srv5 kernel: (ips0) Reset Request - Flushed Cache Jan 6 14:33:06 dev-srv5 kernel: scsi: device set offline - not ready or command retry failed after host reset: host 1 channel 0 id 5 lun 0 ... ad infinitum... Version-Release number of selected component (if applicable): kernel-2.4.21-4.EL How reproducible: Always Steps to Reproduce: 1. Just install Red Hat on an e325 with ServeRAID 2. 3. Actual Results: Loops on disks Expected Results: Works Additional info: This is a more detailed report of the problem: - Software : OS :Redhat AS 3 with updates - Hardware : eseries 325 with 2 internal harddisks (has an LSI scsi raid ctrl onboard) Serveraid 6M raid ctrl EXP400 external storage --> 7 disks/machine used and in non cluster config The problem : The disks can not be accessed with the AMD64 ips driver. This can be found in the logs : Jan 6 14:30:30 dev-srv5 kernel: scsi1 : IBM PCI ServeRAID 6.11.07 Build 2224 <ServeRAID 6M> Jan 6 14:30:36 dev-srv5 kernel: (ips0) Reset Request - Flushed Cache Jan 6 14:30:56 dev-srv5 kernel: scsi: device set offline - not ready or command retry failed after host reset: host 1 channel 0 id 0 lun 0 Jan 6 14:30:56 dev-srv5 kernel: Vendor: IBM Model: SERVERAID Rev: 1.00 Jan 6 14:30:56 dev-srv5 kernel: Type: Direct-Access ANSI SCSI revision: 02 Jan 6 14:31:02 dev-srv5 kernel: (ips0) Reset Request - Flushed Cache Jan 6 14:31:10 dev-srv5 kernel: Error 13 while decompressing splash screen. Jan 6 14:31:22 dev-srv5 kernel: scsi: device set offline - not ready or command retry failed after host reset: host 1 channel 0 id 1 lun 0 Jan 6 14:31:22 dev-srv5 kernel: Vendor: IBM Model: SERVERAID Rev: 1.00 Jan 6 14:31:22 dev-srv5 kernel: Type: Direct-Access ANSI SCSI revision: 02 Jan 6 14:31:28 dev-srv5 kernel: (ips0) Reset Request - Flushed Cache Jan 6 14:31:48 dev-srv5 kernel: scsi: device set offline - not ready or command retry failed after host reset: host 1 channel 0 id 2 lun 0 Jan 6 14:31:48 dev-srv5 kernel: Vendor: IBM Model: SERVERAID Rev: 1.00 Jan 6 14:31:48 dev-srv5 kernel: Type: Direct-Access ANSI SCSI revision: 02 Jan 6 14:31:51 dev-srv5 login(pam_unix)[1195]: session opened for user root by LOGIN(uid=0) Jan 6 14:31:54 dev-srv5 kernel: (ips0) Reset Request - Flushed Cache Jan 6 14:32:14 dev-srv5 kernel: scsi: device set offline - not ready or command retry failed after host reset: host 1 channel 0 id 3 lun 0 Jan 6 14:32:14 dev-srv5 kernel: Vendor: IBM Model: SERVERAID Rev: 1.00 Jan 6 14:32:14 dev-srv5 kernel: Type: Direct-Access ANSI SCSI revision: 02 Jan 6 14:32:20 dev-srv5 kernel: (ips0) Reset Request - Flushed Cache Jan 6 14:32:40 dev-srv5 kernel: scsi: device set offline - not ready or command retry failed after host reset: host 1 channel 0 id 4 lun 0 Jan 6 14:32:40 dev-srv5 kernel: Vendor: IBM Model: SERVERAID Rev: 1.00 While booting the machine will wait till the driver is loaded. After a while (approx 1h) this stops and the disks are still not recognised. The strange thing about the logs is that the separate drives are mentioned while all these drives are configured as 1 logical drive. Also the driver seems to also want to try the channel 0 while this channel is not used (all drives in exp400 are external). Note : The 32bit version of Redhat AS3 on these machines does work perfectly with all controllers and drives. This would be an option if everything else fails but we would like to avoid this. dmesg on the 32 bit redhat version shows correctly the scsi ctrl and EXP400 as the device (not the separate drives).
The ips driver is updated in U1 from 6.00.26 to 6.10.52. The new driver adds support for AMD64 and should solve your problem. Please try U1 (it is expected to ship very soon) and let us know if it solves the problem.
Let me add that the problem is easy fixable by using the driver disk that is included on the latest IBM ServeRaid CD. You can find this at: http://www-306.ibm.com/pc/support/site.wss/document.do?lndocid=MIGR-495PES or http://www-306.ibm.com/pc/support/site.wss/document.do?lndocid=DRVR-MATRIX Since it is not that easy to extract the binary module, it's adviced to rebuild the included sources using the attached Makefile and kernel-ver.c program for your specific kernel. Just untar the ips tarball from the IBM ServeRaid CD into a *new* directory, add the Makefile and kernel-ver.c to this directory. Make sure you have the kernel-source package installed for the kernel you want to build for, and type: make clean all install or to specify a different kernel make clean all install KERNEL_VER="2.4.xx-yyy" Prebuild kernel-module packages will be soon available from: http://dag.wieers.com/packages/kernel-module-ips/ PS Since I'm not able to attach files here (complains about user, and when relogon, complains about No files attached) I'm linking to my website: http://dag.wieers.com/attic/ips/
The latest driver included on the x86_64 (AMD64) U1 CD does not work with this ServeRAID adapter. During installation the module is loaded and the following lines are printed: (ips0) Flushing Cache. (ips0) Flushing Complete. scsi : 2 hosts left and the extra disks are _not_ attached and can't be seen/used. After rebooting the server, the ips driver does work with proper output: (ips0) Flushing Cache. (ips0) Flushing Complete. scsi : 1 host left. scsi1 : IBM PCI ServeRAID 6.10.52 Build 563 <ServeRAID 6M> ... The latest driver includes this changelog for the 6.x series: /* 6.00.00 - Add 6x Adapters and Battery Flash /* 6.10.00 - Remove 1G Addressing Limitations /* 6.11.xx - Get VersionInfo buffer off the stack ! DDTS 60401 /* 6.11.xx - Make Logical Drive Info structure safe for DMA DDTS 60639 And for reference, the output with the latest driver is: (ips0) Flushing Cache. (ips0) Flushing Complete. scsi : 1 host left. scsi1 : IBM PCI ServeRAID 6.11.07 Build 2224 <ServeRAID 6M> Thus we're unable to install Red Hat Advanced Server on an e325 with a recent ServeRAID card. (But we can install it on the onboard scsi controller using module 'mptscsih' though) Use the procedure described above to compile your own ips module.
An errata has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on the solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHSA-2004-017.html