H165929: ECA029: REPLACEMENT OF NETFINITY FIBRE CHANNEL CONTROLLER PURPOSE: This ECA provides for the replacement of the Netfinity Fibre Channel RAID Controller (Standard Controller shipped in the 3526) and the Netfinity Fibre Channel Failsafe RAID Controller. Product evaluations at IBM have revealed that you may experience Fibre Controller hangs, reboots, a failed controller and possibly incorrect data with or without an error message when using the Netfinity Fibre Channel RAID Controller and/or the Netfinity Channel Failsafe RAID Controller. Customers are advised to replace the Netfinity Fibre Channel Controller, and, if installed, the Netfinity Fibre Channel Failsafe RAID Controller as soon as possible. This is a MANDATORY ECA. FEATURES: Type, Machines Affected Model, With and/or Feature/Device B/M to be Service System Stage B/M Description Installed Hours Hours 3526 B/M10L6993 FIBRE CHAN. CONTROLLER B/M37L0412 00.5 00.3 PHYSICAL CHECK: - The system is any of the following IBM servers: a Netfinity 7000-M10 server, Type 8680, any Model. a Netfinity 7000 server, Type 8651, any Model. a Netfinity 5500-M10 server, Type 8661, any Model. a Netfinity 5500 server, Type 8660, any Model. - The system is connected to a Netfinity Fibre Channel RAID Controller Unit, Type 3526, Model 1RU or 1RX. - There is not a visible FRU Label on the front of a suspect controller or the Failsafe RAID Controller when the controller is fully seated in the 3526 unit. A good controller will have a FRU Label (FRU p/n37L6077) visible from the front, when the controller is fully seated in the controller unit. PREREQUISITES: If both the controllers in the 3526 are identified from the physical check above, as needing to be replaced, you must have two (2) new FRU p/n37L6077 controllers to replace both. You cannot replace one now and one later, they both have to be done at the same service call. COMPANION: NONE CONCURRENT: NONE DETAIL: Replace the Netfinity Fibre Channel RAID Controller (FRU p/n10L6993), and if installed, the Netfinity Fibre Channel Failsafe RAID Controller (FRU p/n10L6993) with FRU p/n37L6077. Instructions on replacing the Netfinity Fibre Channel RAID Controller (FRU p/n10L6993), and if installed, the Netfinity Fibre Channel Failsafe RAID Controller (FRU p/n10L6993) with FRU p/n37L6077. 1. Identify the mode of the controllers. a. Open SYMplicity Maintenance and Tuning application b. Click on Options c. Click on Controller Mode d. If Controller Mode is grayed out (i.e. you cannot click on it) write down mode=Active Active. If you can click on it, write down which controller is Active (A or B). 2. Ensure that the server is using the latest device driver for the Netfinity Fibre PCI Adapter. Note: These drivers can be found at the following URL: http://www.pc.ibm.com/searchfiles.html search on "Fibre Drivers" If they are not using the latest device driver, upgrade to the newest device driver. Pages 8 & 9 of the "Netfinity Fibre Channel PCI Adapter Software Installation Guide" describes the procedure for Updating the currently installed Windows NT Driver. Note: The "Netfinity Fibre Channel PCI Adapter Software Installation Guide" is available in .pdf format from the following URL: ftp://ftp.pc.ibm.com/pub/pccbbs/pc_servers/24l8026.pdf 3. Stop all applications and down and power off the servers. If in cluster environment be sure to take down all servers, one at a time. 4. Power off the 3526. Note: Be sure to turn off the 3526 using the two (2) power switches on the back of the 3526. 5. Remove the old Netfinity Fibre Channel RAID Controller(s). Note: If there is a second (failsafe) controller installed, and it also meets the ECA identification standards set forth in the Physical Check section of this ECA (i.e. lack of FRU label visible when fully seated) then both controllers must be replaced at the same service call. 6. Once the Controller(s) have been replaced, power on the 3526 (two (2) power switches on the back). Important: You must wait at least 2 minutes before powering on the servers. It takes 2 minutes for the controller to come up. 7. After waiting at least 2 minutes, power up the servers as you would normally. In a clustering environment this would mean bringing up one (1) server Node at a time. 8. Once the servers are up and running do the following: a. On a server, open up a DOS window. b. Change directory to c:\program files\symsm c. Type "clean -all" and press return. d. Close the DOS window. e. Open SYMplicity Maintenance and Tuning application f. Repeat steps a-e on all remaining servers connected to the 3526. 9. Now ensure that the controllers are set to the same mode as was identified in Step 1. a. Open SYMplicity Storage Manager b. Click on Options c. Click on Controller Mode d. If Controller Mode is grayed out, then continue on to the next step. Note: If it is not grayed out, and was before, then click on the Active/Active button on the bottom left of the screen. Otherwise it is in an Active Passive mode, and ensure the same controller that was Identified as Active in Step 1d is also identified as active now. If it is not then press the Switch Active/Passive button in the lower middle of the window. 10. In addition once the controllers have been replaced and the system is functioning correctly, do the following to determine if there has been any incorrect data written: This is done by running the Manual Parity Check/Repair option in the Recovery Application outlined in the IBM Netfinity SYMplicity Storage Manager User's Handbook. The highlights of this are given below for convenience. Running Manual Parity Check 1. Use the following procedure to run Parity Check/Repair manually. 2. Start the Recovery application. 3. Select the RAID Module containing the LUNs you want to check (or select ALL RAID Controllers). 4. Click the Manual Parity Check/Repair button or select Options -> Manual Parity Check/Repair from the drop-down menus. 5. After you have selected all the LUNs to check, click Start Parity Check/Repair. 6. As each LUN is checked, a histogram bar appears on the screen indicating the Parity Check/Repair progress on that LUN. 7. When Parity Check/Repair is completed, you will see a message indicating if any errors were found. Notes: - You can not run Parity Check/Repair on a RAID 0 logical unit or a non-optimal RAID-1, RAID-3, or RAID-5 logical unit. - Parity repair fixes parity, not data. If the parity inconsistencies were caused by corrupted data, the data is still corrupted, even though the parity is correct. - While parity check/repair is in progress, you can not perform other Recovery tasks. TRADEMARKS: Other company, product and service names may be the trademarks or service marks of others. NOTES: This ECA is "MANDATORY". This ECA is scheduled to be withdrawn January 31, 2000. Parts in "Features" Section, under "B/M to be Installed" column, MAY BE Ordered CODE "A". IF Machine is down, AND B/M37L0412 is not available, THEN FRU p/n37L6077 can be ordered to get machine back up. USA: IBM PSS/TSS CE's should record all time and parts to Service Code 33, ECA029, Other Office 990. Travel Time = 0.8Hrs. EMEA: IBM PSS CE's should record all time and parts to Service Code 33, ECA029, (M/T number) Other Office 990. Travel Time = 0.8Hrs. EMEA Business Partners: Refer to Warranty Claim System; Use Emergency Claim 5 (ECA) and enter into CPPS: ECA Number 029 Machine Type as required (number). DATE: February 22, 1999