From: faq-rt@antimatr.hou.tx.us (Mark Whetzel) Newsgroups: comp.sys.ibm.pc.rt,comp.unix.aix,comp.answers,news.answers Subject: COMP.SYS.IBM.PC.RT: AIX V2 FAQ - Question/Answer, Part 1 of 2 Asked Questions (and their answers) and general information, hints and tips on issues related to using AIX V2.2.x on IBM RT (Model 615x) computers. Keywords: IBM RT PC software FAQ 6150 6151 6152 AIX VRM AOS MACH PICK Message-ID: Date: 5 Jul 94 03:57:33 GMT Expires: 17 Aug 94 03:56:49 GMT References: Sender: markw@antimatr.hou.tx.us Reply-To: aix-rt@antimatr.hou.tx.us Followup-To: comp.sys.ibm.pc.rt Lines: 1108 Approved: news-answers-request@MIT.Edu Supersedes: Xref: news.lth.se comp.sys.ibm.pc.rt:2924 comp.unix.aix:41904 comp.answers:6164 news.answers:24717 Archive-name: ibm-rt-faq/aix-v2/misc/part1 Last-modified: Fri Mar 25 03:37:40 CST 1994 Version: 1.1 Frequently Asked Questions for: AIX V2.2.1 on IBM RT systems Various subject Q & A topics for AIX/RT. Part 1 of 2 parts. The AIX V2 FAQ currently consists of EIGHT parts in 5 topic areas: AIX V2 FAQ Index and acknowledgements AIX V2 OS Specific hardware issues - in one part (currently). AIX V2 Software questions - in three parts. (This FAQ) AIX V2 Problem Questions and answers - two parts. AIX V2 Porting software notes - in one part (currently). See the AIX V2 FAQ index posting for a complete index of the AIX FAQ postings. This FAQ covers geneal questions and answers to using AIX V2 on an IBM RT system. It also contains some general HOW-TO questions and help. ================================================================================ 3.0 Problem Question and Answers. Q.1 I have lost my root password, how can I reset it? I have a procedure in another document outlining the steps for this recovery. E-mail me with your request, and I will mail it to you. I hesitate to post the procedure to a public list. ------------------------------ Q.2 LED code 99 is displayed at power on, what is it? The keylock has been left in the 'locked' position at power on. Turn the key to unlock and the system will continue booting. ------------------------------ Q.3 - How do I move 'system' filesystems (/,/usr,/tmp,/u,/vrm)? How to increase the size of filesystems. The procedure for moving or enlarging a minidisk are virtually the same. The basic procedure involves rebooting from the install/maint disk, backing up the minidisk in question, deleting the original space and remaking the new space, then restoring the data. One critical element of these procdures is the IODN of the various system minidisks. Make sure that you use the same IODN number for the newly created area as before. * Enlarging an existing minidisk To enlarge a minidisk, the principle is the following: backup your minidisk load the installation/maintenance diskette get the IODN of the current minidisk you want to enlarge delete the minidisk create a new, larger minidisk using the same IODN make a filesystem on the new minidisk restore your data reboot normally * Detailed procedure: You can change the size of the /usr minidisk (or any other for that matter). Minidisks are sections of the hard disk that are contiguous. There may not be any free space immediately after the /usr minidisk on your system. If that is the case, then you will have to rearrange several minidisks on your system in order to create the amount of contiguous free space that you need. A general procedure is outlined below. More detailed information is contained in Managing the AIX Operating System (SBOF-0168). 1) While in the Bourne Shell use the minidisk command and print off the minidisk configuration (F4). 2) Check the printout from the minidisk command. Check to see if there is any free space immediately after the /usr minidisk. If there is enough, then the procedure will be fairly simple. If not then it is a little more involved. 3) Shutdown and boot from the Installation/Maintenance Diskette. Follow procedure A or B, depending on the free space. Procedure A - IF there is ENOUGH free space: 4) Select Backup commands. 5) Select Backup a file system and backup the /usr minidisk. DO NOT USE THE BACKUP A MINIDISK IMAGE OPTION. Be sure to check for any errors during the backup operation. 6) Select Delete a fixed disk minidisk and write down the /usr minidisk IODN. Delete the minidisk. 7) Select Create a fixed disk minidisk. a) Select Specify IODN and supply the IODN that you noted in the previous step. b) Select No preference to cause the minidisk to be created in the first disk area large enough to accommodate it. (Do not use the beginning, middle, or end allocation suboptions.) c) Specify the number of blocks necessary to meet your new requirements. 8) Select Make a file system and create a file system on the new minidisk. Select the same IODN that the original file system had, the one you noted above. 9) Select Restore commands. a) Select Restore a file system. b) Restore the old file system to the new minidisk, using the same IODN. 10) Select Check a file system and check the structure of the new file system. 11) Remove the AIX Operating System Installation/Maintenance diskette and start the system from the fixed disk. Procedure B - IF there is NOT ENOUGH free space: 4) Select Backup commands. 5) Select Backup a file system and backup the /usr filesystem. a) Repeat this step for the other filesystems that are on the hard disk and physically located "behind" the /usr minidisk. For example if a system only has one hard disk (70 Meg), the minidisk facility gives the following information: MD MD MD Block Number ... Mount ...... Name IODN Type Size Blocks ... Directory ...... hd6 32760 VRM 512 3600 /vrm 32766 Pgsp 512 13722 hd0 16384 AIX 512 25679 / hd2 16387 AIX 512 66809 /usr hd1 16386 AIX 512 22810 /u hd3 16388 AIX 512 4000 /tmp hd4 16385 AIX 512 5000 (the dump minidisk) Available space 512 1000 Assuming that there is some free space after hd4, then backup the /usr and /u filesystems. You probably would not backup the /tmp minidisk since you do not have any permanent files there. You would not backup the dump minidisk (IODN 16385). DO NOT USE THE BACKUP A MINIDISK OPTION. It is good backup procedure to check the file system using fsck to make sure it isn't corrupted before doing a backup. Also, make sure the backup was successful by doing a 'restore -T' to the backup tape, before destroying the file system. 6) Select Delete a fixed disk minidisk and write down the IODN numbers for every minidisk you are about to delete. From the example in step 5, you would record the IODN's for the /usr, /u, /tmp, and the dump minidisk. After verifying your tape(s) are good, delete the minidisk(s) that need to be re-arranged. 7) Select Create a fixed disk minidisk. a) Select Specify IODN and supply the IODN that you noted in the previous step. b) Select No preference to cause the minidisk to be created in the first disk area large enough to accommodate it. (Do not use the beginning, middle, or end allocation suboptions.) c) Specify the number of blocks necessary to meet your new requirements. d) Repeat step 7 a), b), and c) for each of the minidisks previously removed. 8) Select Make a file system and create a file system on each of the new minidisks. Be sure to use the same IODNs as was used on the original file systems. 9) Select Restore commands. a) Select Restore a file system. b) Restore the old file system to the new minidisk, using the same IODN. c) Repeat 9 a) and b) for each minidisk and file system. 10) Select Check a file system and check the structure of all the new file systems you have restored. 11) Remove the AIX Operating System Installation/Maintenance diskette and start the system from the fixed disk. ==**==**==**==**==**==**==**==**==**==**==**==**==**==**==**==**==**== Tip from Willi Zuber, IBM France Date: 16.09.1988 [ NOTE: I personally have NOT tested this way of 'increasing' /usr. I HIGHLY recommend that you have a good system backup in case something goes wrong. Always 'Be Prepared' when experimenting! - MW ] If /usr is full, do the following to gain some space: 1. Find the biggest subdirectory in the /usr directory: commands: cd /usr li -R1p | xargs -i du -s {} The system lists the directories of /usr and their respective disk usage. Take note of the biggest one: ex. 18000 blocks for /usr/lib [ Note: lib is the hardest to move, as it is usually busy with programs. I suggest /usr/lpp first, then try /usr/lib. I suspect it would be best to have a completely idle system and a 'killall' done first before attempting /usr/lib. -MW ] 2. Create a new minidisk to contain that "biggest subdirectory" and give it a "dummy library name": ex. hd13 with 22000 blocks to contain "/dummylib" 3. Mount the new minidisk: command: mount /dev/hd13 /dummylib 4. Kill the error daemon: command: kill /usr/lib/errdemon NB. This is only necessary if you move the subdirectory "lib" 5. Copy /usr/lib to /dummylib: commands: cd /usr/lib find . -print | cpio -pdl /dummylib [ Note: Verify that your copy is good before continuing. - MW ] 6. Umount /dev/hd13: command: umount /dummylib 7. Remove the entries of /usr/lib: commands: cd /usr/lib rm -r * 8. Mount hd13 on the correct directory: command: mount /dev/hd13 /usr/lib 9. Update the file /etc/filesystems: change the stanza name "/dummylib:" into "/usr/lib:" (do not forget the double point!!) [ Colon for us USA types - MW ] WARNING: Make sure that the mount is set to always mount at IPL. 10. Re-ipl the system. ------------------------------ Q.4 - What do the IODN numbers mean. Input Output Device Numbers (IODN) of the system minidisks. The numbers assigned to various system minidisks is how the VRM is able to associate a particular disk partition with the related filesystem. dev IODN use mount attributes --- ----- ------- --------------- ----------------- hd0 16384 AIX / Auto IPL hd1 16386 AIX /u hd2 16387 AIX /usr hd3 16388 AIX /tmp hd4 16385 dump 32766 PgSpace hd6 32760 VRM /vrm read only hd7 20001 AIX /user-minidisk IODN number for minidisks is 16384 (minimum) to 32767 (maximum). User created minidisks start at 20001 thru 29999. ------------------------------ Q.5 - How do I make a netnews filesystem with lots of inodes? The minidisks command can make a minidisk partition of any size, but the maximum number of inodes in will make is less than 32000 inodes, and is based on the size of the minidisk being created. The AIX filesystem can be made with a larger number of inodes by use of a prototype file and the 'mkfs' command. 1) Use minidisks to make a filesystem of the size you wish, on the disk you want.. example /news. number of blocks maximum on a E310 = 605460 (leaving 9 free) Note the device number when the minidisks makes the new filesystem and minidisk. 2) mount /news df /news Note: The number of blocks will be LESS than that specified in the minidisks command, because of inode blocks overhead. 3) Use the proto command and save the output to a file. proto /news >/tmp/proto.keep 4) unmount /news 5) Edit the /tmp/proto.keep file and add the size of the filesystem, and your requested number of of inodes. Be conservitive, the 'mkfs' command will round UP to a proper multiple based on the geometry of the target disk drive. # cat /tmp/proto.keep : Prototype file for /news /noboot 605460 63000 d--777 0 0 lost+found d--755 0 0 $ $ $ # mkfs /dev/hd10 -vNEWS /tmp/proto.keep DANGER: Do not go over 64000 inodes. Due to some bug in the AIX filesystem code, it seems to get lost if the total number of inodes is very large, and does not contain some slack space. 6) mount /news Expand the lost+found directory by creating about 200 files with touch, and then delete them. This expansion is required so that during 'fsck' processing, if errors DO occur, the system can move 'dangling' files to the lost+found directory. During this processing fsck cannot have the filesystem changing without its knowledge. ------------------------------ Q.6 - The domain name server is not working correctly. The version of named shipped with AIX/RT is VERY old! A partial 'what /usr/lpp/tcpip/etc/named' shows: ns_init.c 4.23 (Berkeley) 2/28/88 ns_main.c 4.30 (Berkeley) 3/7/88 Copyright (c) 1986 Regents of the University of California. ns_maint.c 4.23 (Berkeley) 2/28/88 ns_req.c 7.7 89/01/27 11:30:46 ns_req.c 4.32 (Berkeley) 3/31/88 ns_resp.c 4.50 (Berkeley) 4/7/88 ns_sort.c 4.3 (Berkeley) 2/17/88 ns_stats.c 4.3 (Berkeley) 2/17/88 named 4.8 #12: Wed Jul 27 11:47:05 CDT 1988 While it pretends to be 4.8, it does not have anywhere near the features of the 4.8.3 server or later. Particular problems I have seen are: - Secondary servers do not understand how to save the transfered zone data to files for restart recovery. - zone transfers are NOT asyncronous. If lots of secondary zones are being handled, zone transfers will tie up the server making in unresponsive. - 'TXT' records, while not documented as available in the TCP/IP manual, do work on IBM supplied server. - Later maintence level of 1773, seems to have broken MX record lookup by the resolver routines located in libc. If you are having problems with the server, I suggest that you replace the named server with at least the 4.8.3 version available from the net. See section P.3 on porting and running that server. ------------------------------ Q.7 - Mail addressing and setup. The AIX/RT standard mail facilities consist of the sendmail program, (if installed) and the SYSV based user agent 'mail' command. The sendmail code appears to be based on sendmail version 7.24. The only overt major change to the programs is the location of the control files. They are located in: /usr/adm/sendmail IBM includes a program called '/usr/lib/edconfig' to help with editing the sendmail.cf control file. This program is menu driven, and depends on the supplied sendmail.cf and the comments located in the file to be able to change the sendmail configuration for you. However, you do not have to use this program to configure the sendmail config file. There are liberal comments located in the file describing the various macros for the system configuration. For more documentation on the IBM sendmail, see the "Managing the AIX Operating System" manual number SC23-2008. Chapter 7 is devoted to mail issues. That chapter has some excelent information on understanding the sendmail.cf file, sendmail macros, mail tailoring issues, understanding mail rewrite rules and the obscure format of the sendmail config file. Also covered is the format of the queue control files, and more. If you use the supplied AIX sendmail config file, here are some macros that should be looked at: Dw Host name macro Cw Host name class DD Domain name macro DE Macro for first (left most part here in the US) domain part. DF Macro for second domain part. DG Macro for third (optional) domain part. DH Macro for fourth (optional) domain part. To avoid confusion IBM manual recommends that the following names for your system should all be the same: nodename - Defined in /etc/master and built into the kernel. Binary changed in /unix and the running kernel with the 'chparm nodename=' command. nickname - The name that Distributed Services uses for your system name. Changed by the 'ndtable' command. hostname - TCP/IP name defined and displayed by the 'hostname' command. Usually set from /etc/rc.tcpip shell script. As supplied, you will find that the IBM mailer has a great deal of problems existing in a mixed UUCP and TCP envionment. A pure TCP envrionment also will have problems with MX records, and domain name service queries. X.400 names can cause it all kinds of problems. I found that the best solution was to replace the supplied sendmail with PD versions. See section P.5 for porting and install instructions for IDA sendmail 5.65c. ------------------------------ Q.8 - How do I set up a dial-in autospeed detect serial line? See the software question S.13 for general serial port add instructions. Modem capabilities vary greatly from vendor to vendor, and much of the 'art' of serial port setup of any type will be impacted by the setup 'switches' both hardware and software register setups. A great deal of the problem with serial port function is using the proper cable. Modem usage requires a correctly wired cable. See the hardware FAQ for pinout and cable information. Before starting any attachment of a modem, read your vendors manual on setups. Some vendors documentation have recommended setups for attachment to UNIX systems, particularly when used with UUCP. Examine your vendors modem manual about soft register settings dealing with speed matching, and how it is presented to the host computer. If you have one of the newer breed of modems where the serial-port-to-LOCAL-computer speed can be locked to a single speed, while the modem-to-REMOTE-computer speed can be set to the callers speed, Then the procedure described below is not neecessary, simply setup the port as normal, and set the port speed to the 'locked' speed of the modem, and your work is done. (Trailblazer modems have this capability for example). Make sure that the pacing, either XON/XOFF or RTS/CTS, is setup correctly in the modem and/or the communication terminal that the caller will be using. If the users speed will be mis-matching the port speed, there can be over/under run problems. Setting up an AIX/RT port for autobaud capabilities is not described very well in any of the AIX documentation. The Managing the AIX Operating system book has a small section devoted to serial port setup, but this obscure feature of AIX/RT serial ports is overlooked. The real truth is AIX/RT serial port capabilities are not great, especially in the autobaud arena. You will have to decide what combination of speeds and parity settings you would like to support and select a few popular combinations. To allow more than one speed to dial in requires you to edit manually the /etc/ports file AFTER you add the tty port with the devices command. WARNING: Any use of the devices command on this port later will wipe out the following changes, and will have to be manually be re-entered for that port. Edit the /etc/ports file, find the /dev/tty## entry that was added by the devices command for your serial port. Looks a little like this: /dev/tty0: lock = true speed = 2400 logmodes = echoe+cs8 runmodes = hupcl+cs8+cread+brkint+icrnl+opost+tab3+onlcr+isig+ icanon+echo+echoe+echok parity = none synonym = /dev/ltty0 Change the 'speed' and parity entries to look like this for a three speed call in of 9600,2400,1200 with no-parity used for any of the speeds used. speed=9600,2400,1200 parity=none,none,none This will set the port initially to 9600 baud, and step down the speed on each recipt of a 'break' sequence by the modem. If a user dials in at 2400, after the modems sync carrier, and setup for 2400 baud, if the port is at the wrong speed, they will receive garbage for the 'hearld' stanza in the 'defaults' section of the /etc/ports file. They then press 'break' and the RT will move to the next speed, and then the 'hearld' will be re-displayed. Once the speed and parity setting matches the users modem speed, the login hearld will be readable. If the end of the list is reached on speeds, it wraps around to the first entry. ------------------------------ Q.9 - How to setup a serial port for shared dial-in and out? To properly allow a serial port to be used for dial-out with the cu command or for UUCP usage, AND allow call-in capability will requre a few non-standard, and obscurely documented changes. 1) Define the port for dial-in usage as described in section Q.8 above, but make sure the 'ae' parameter is set to delay or share. I find that delay works best, but use what is proper for your requirements. 2) Add the following line to the /etc/ports file for the /dev/tty# stanza requiring dial-in/out capability. lock = true This will cause a lock file to be created in /etc/locks by the getty, and also allow it to correctly handshake with uucico and cu to release the serial port when necessary for outbound operations, and also will block uucp and cu from operating when a caller is logged in on the port. Under normal usage, this is all that should be necessary. NOTE: Even on the last fix level, the ATE command does not know how to request a lock, or release a port in a automatic fashion. You will have to issue a 'pdisable' command prior to 'ate' usage, and then re-issue a 'pdelay' or 'pshare' command after done using ate. NOTE2: Proper locking and releasing of ports with uucico and cu was not fixed until around the 2705 opsys update level or later. ------------------------------ Q.10 - The devices command won't find my newly installed adapter. Four possibilities: 1) Conflicting interrupt or DMA levels are preventing the board from being properly recognized at boot time. 2) The adapter card is not supported by AIX V2.2.1 3) The adapter card is defective. 4) Needed support functions were not installed. The VRM device driver LPP sections may not be installed. The disks contain support for the following drivers: IBM SCSI cards, 3270 cards, Ethernet cards, Token-ring cards, devices needing block I/O support. ------------------------------ Q.11 - What is the procedure to add a hard disk to a system that has been in use on another RT system running AIX. Assming that the drive to be added came from another RT system, no special formatting is needed. The minidisk partition table, and bad track map are already on the drive. To be safe, I personally like to remove all hard drives except the NEW one, put the terminator on the new drive, boot from the VRM floppy, and then use the second procedure below to remove any stray minidisks from the foriegn drive. Simple procedure: *IF* you are sure that the drive to be added had NO minidisks on it, all you have to do is: 1) Plug in the new drive, observing that the terminator is removed. Only one disk must have a terminator, with one controller card, If you have a system with more than one controller, follow the cable set for each controller. Only one drive for each set must have a terminator resistor installed. 2) Use the 'minidisks' command to manage the drive, add minidisks, filesystems, and enjoy. Your work is done. Complex procedure: However... If the new drive had minidisks already on it, and worse, SYSTEM minidisks on it, AIX and VRM will get VERY confused about what minidisks are what. Use this procedure if you are not sure what minidisks may be on the drive, if any. This is the ONLY safe way to check out foriegn disks. Don't try to boot your AIX system just after adding it to the system. 1) Boot the diagnostics disks, and see what drives are found. Use the option 'show all installed options' selection of the utilites portion of the diagnostics. Make sure that the terminator is installed correctly as shown in the hardware setup FAQ or manuals. Remember, only the last drive in the chain for each controller just have a terminator pack. 2) I like to test the drive with diagnostics and make sure it is at least responding to initial selection. 3) Boot up the VRM diskette #1 and use the VRM maint utility commands 'qi' and its other very strange commands to display all minidisks on each and every drive. Write down the IODN and IOCN numbers on each is installed drive. If the NEW drive has any VRM minidisk numbers (32760 and 32766) AND AIX system minidisks (16384 and up to 19999). VRM will get VERY confused about what VRM and AIX filesytems are what. It seems that VRM goes by the IOCN/IODN numbers to figure out what partitions are what, and gets very confused if there are duplicate partition numbers. To clear this up continue with step 5. 4) If ALL the minidisks are over 20000 and less than 32760 in number, AND *NONE* of the minidisks has the 'PgSpace' marker, you can do one of two procedures: A) Exit the VRM utilites and re-boot your normal AIX system, then use the 'mdrc' command to recover any filesystems and minidisks that were on the 'new' drive. The mdrc (minidisk recovery) command reads the minidisk partition table, and re-writes it and resyncronizes the /etc/system and /etc/filesystems files so that the numbers on it and the system tables will not conflict with existing minidisk numbers. This procedure will rename any mount points on the 'new' filesystems to /tmp/directory/hd## matching the renamed 'hd' numbers that will be assigned to the added drive. B) Use the VRM install utilities and use the delete minidisk comand descrived in step 5A to delete any or all unneeded minidisks. If any minidisks are left, then use step 4A on the disk to recover any remaining minidisks, and syncronize the system tables. 5) If some or all the minidisks are under 20000 in number OR there are minidisks in the 3276x range, OR one of the minidisks has a VRM 'PgSpace' marker, problems can occur. You can choose several options: A) Delete everything on the new drive, B) Delete some minidisks and recover what is left, or C) attempt to recover all minidisks, including the 'system' ones. While following these proceedures, BE CAREFUL on what drive is called 0 and 1 and 2. Drive 0 normally is the top back drive connected to the C1 connector. Drive 1 is the bottom back drive connected to the D1 connector, and drive 2 is the front drive bay connected to the E1 connector. Things can get VERY confused if you have a system with two controller boards in slots 1 and 2. A mis-cabling by some other person can scramble the drive numbering. A) You can remove all the minidisks on the drive by using the delete minidisk command 'dlmd', to delete any of the possibly duplicate minidisk(s). If any to be removed are a pagespace, follow the procedure in step 5C2 below prior to deleting that type of minidisk. The syntax of the delete command is: dlmd minidisk_id = IODN or name of minidisk being deleted. reald = The fixed disk IODN on which the minidisk resides. If you have more than one minidisk with the same name on different drives you *MUST* specifiy this parameter. The number to be used is the disk drive NUMBER (0,1 or 2). Once all old minidisks are gone, remove the VRM utility disk, re-install your other disks (if you removed them), and reboot AIX and use the now empty disk as normal. B) Using the 'dlmd' command as described in procedure 5A above, delete any system minidisks from the new drive, leaving the 'user' minidisks (greater than 20000 and less than 32767). Continue with step 4A above to recover the filesystems. C) To recover ALL minidisks for examination, use the VRM utilities to remove the special VRM markers from the minidisks on the new drive. There are three special case filesystems: 1) If any have the 'PgSpace' marker, it is a paging area, and is most likely currently in use by this boot of VRM (from the utilities!). Deleting it now can cause a run error. First, unmark the minidisk as a pagespace with the utility command 'setattr'. Reboot the VRM utilities, and then remove any minidisk(s) that were in use as pagespace (there is no AIX filesystem on it and nothing generally of use anyway). 2) If any minidisks have the IODN of 32760 and have the 'VRM' marker, AND you want to recover it as a regular filesystem minidisk, remove the VRM marker from the minidisk with 'setattr'. You can also use the setattr command to remove the read-only status of the minidisk. 3) If the '/' (root) filesystem exists with IODN 16384 on the drive, remove the 'IPL' marker from the minidisk with 'setattr'. Then continue with step 4A above using the AIX/RT 'mdrc' command to renumber and rename the remaining possibly duplicate minidisks. ------------------------------ Q.12 - The disk I have is one of the IBM supported disks, but it can't be found by diagnostics, or AIX. The most likely cause of this problem is the IBM supplied minidisk partition table, and the defect map area has been removed by some method. A full reformat with a non-AIX/RT computer is the usual cause of this problem. To add this disk, follow the procedures described in Q.13 below for adding a ODM ESDI drive to the system. When you get to the part on specifying the disk parameters, use the option '1 - Specify a known device.' to fill in the disk parameters. ------------------------------ Q.13 - What is the procedure for adding a Non-IBM ESDI or MFM hard disk with AIX V2. To use a NON-IBM ESDI drive on an RT, it will have to be very close in its geometry to one of the IBM supported drives. I have not yet tried to use ESDI drives with larger than 310 Mbytes of space with AIX. To be able to properly use the drive, it will have to be prepared in two seperate formatting steps. First, shutdown the system as normal. I personally like to work on new disk drives with NO other drives installed in the system. The serves several purposes. 1) If there is anything wrong with the drive, it won't cause problems the other disk drives in the system. 2) If you make an error specifiying any arguments or commands with the diagnostics or the VMF utilities, you won't accidently damage any of the information on your other drive(s). The RT employs a specially cut and twisted drive control cable, allowing all hard drives to be jumpered for drive select 2. Examine the disk drive electronics, and set the drive select jumpers for the drive to DS2. Double check other jumpers if necessary on the drive. Some drives have jumpers to control the number of sectors per track, and timing adjustments. Look at the ESDI drive chart in this FAQ and pick a setup that is close in sectors to the IBM drives. The larger drives usually use from 34 to 36 sectors per track. If you are following my precaution for testing the drive by itself, make sure that the proper value and type of terminator resistor is installed in the drive, as this will be the last (only) drive in the chain of drives in the system. Install the drive in one of the open drive bays. Make sure that the drive is plugged into the proper power cable. The tower model 6150 RT as the following drive bay locations. Drive bay 'C' is the top back full height drive bay, and uses power connector P13. Bay 'D' is the bottom back bay, and uses power connector P14, and bay 'E' is the front full height bay just below the two half height floppy drive bays. If this is the only drive in the system, put it in bay 'C'. Unfortunately the desktop 6151 system has only one full height drive bay. While the controller is capabile of running two (or three depending on the exact model controller) hard drives, I have not investigated using two half height drives for this system. That may be possible as long as you don't exceed the capacity of the power supply. At this point, I like to boot the diagnostics diskettes and see if the drive is recognized. Choose the utility section of the diagnostics, and select the 'show all installed devices'. All adapters in the system should be listed, including the fixed disk drive adapter. This makes sure that the adapter is still functioning. Usually the drive itself will NOT show up in the attached devices list, even if everything is correct with the hardware. The AIX RT diagnostics are rather dumb and will not understand the drive and its geometry if there is no AIX style minidisk partition table on the drive. If the new drive DOES show up, the diagnostics will show its position and size if it is an IBM drive. Unless you would like to completely reformat the drive you should skip to the FAQ section on adding a drive with existing AIX formatting to the system. There may be minidisks on the drive that can conflict with your system. If you do wisk to reformat this drive, skip past the paragraphs below about VRM formatting. We now have to assume that the ESDI drive is working correctly, as none of the AIX or diagnostics tests will show the drive as operable yet. The new drive must be formatted with the VRM installation disks. This formatting will put a new minidisk partition table on the drive, and also will create the VRM bad track map table on the drive (empty), and the geometry information record will be written to the drive. Put disk 1 of the VRM install disks in the floppy drive, and boot the system from the floppy. Use the following procedure to walk thru the menus of the VRM utilities and the VRM format command. When the VRM install disks boot up, they will ask you for your keyboard language, choose the proper menu item and press enter. Next a menu of VRM options will be displayed. * Choose menu option 7, "Maintenance Facility" You will be given a command line interface with a prompt of ">". To get help with any comand type 'help commandname". The screen will not scroll when the screen is full with these utilities. Press F1 to see more of the screen. Assuming that the drive is in bay C, and all by itself, type the following command to start formatting that drive: format 1 Drive bay C is drive 1, bay D is 2, and bay E is 3. Systems with two EESDI controllers thus have drives 1,2,3 on the first adapter and drives 4,5,6 on the second adapter. * A display menu of adapters in the system will now be shown, and the attached drive numbers that may be formatted. At the prompt to continue, press 'Y' VRM will complain that the drive is unreadable, with the following message as there is not an existing partition table on the drive. * Drive is Un-readable. No valid disks parameters can be located. Do you want to specify parameters (y/n). Enter "y" to continue. A menu of selectable drive parameters will now be displayed that looks like this: 1 - Specify a known device. 2 - Specify number of cylinders 3 - Specify number of heads 4 - Specify number of sectors per track 5 - Specify interleave factor yes 6 - Specify cylinder where precomp begins. yes 7 - Specify seek profile. 8 - Specify skew factor. 9 - Display configuration record 10 - Continue with format. 11 - Quit format command. If you have an IBM drive that has had its VRM minidisk table wiped out, or a drive that exactly matches the geometry of an IBM drive, you can choose menu option 1 and a menu of IBM known drives will be displayed. Choosing a IBM drive type will then bring you back to this menu with 'yes' by all of the parameters that have been set. If you have a drive with different specifications, choose each menu option in turn above and enter the proper value for your drive. As each option is specified, and 'yes' will be entered next to that menu option. Note: ESDI drives do not need menu option 6 or 7, that is why the 'yes' is already on those options. When you have finished specifying the settings, menu option 9 will give you a summary of the settings, or allow you to display the current settings that a drive already has. When ready to continue formatting, choose option 10. As the drive formats it will show you the cylinder number in a running display. If there are problems, you will see the following error message, or similar: Start I/O failed. IODN=1315 return code= -6 I/O error during format operation. drive 1 cyl 0 head 0 Do you want to continue format? This error indicates an immediate failure (cyl=0, head=0) and most likely the drive is not responding, or a gross misarrangement of geometry settings. A 'n' will abort the format operation, a 'y' will attemt to continue. A very bad track on a drive can also show this error, but the cylinder and head numbers must be non-zero. A drive with cylinder zero defective is most likey useless. Assuming that the VRM format operation has completed, now you need to boot the diagnostics disks. VRM formatting will now allow the diagnostics to be able to find and understand the drive.. HOWEVER, the bad track map is not filled in by the VRM formatting. After booting the diagnostics, choose menu option two (2), "Utilities". This will bring up the "Utility Selection" menu. Choose option three (3), "Fixed Disk Utilities". Option one (1) on this menu will be "Format a Fixed disk". Make sure that you select the correct disk. Confirmation of format will not be requested. The disk size will be known from the already installed minidisk partition table from the VRM formatting step. After formatting is complete, exit the diagnostics, and remove the diagnostics diskette, and enjoy your new disk drive. ------------------------------ Q.14 - What is the procedure for using non-IBM SCSI disks with AIX V2. As mentioned in the hardware FAQ, the IBM RT SCSI adapter uses a differential interface, instead of a single ended interface like that used in most workstations. This makes it electriclly incompatible with most SCSI drives on market. However, all is not lost. :-) Some vendors do offer their SCSI drives with an optional differential interface. There also exist a couple of manufacturers of 'Differential-to-single-ended' converter boxes. These boxes allow a bi-directional conversion of the SCSI bus and will allow attachment of these single ended disk drives. Newer IBM RS/6000 systems have optional differential interfaces now available, and some SUN systems will be offerd with differential interfaces. This will most likely cause drive manufacturers start to increase the availability of differential drives. Here is a procedure that works for me to attach SCSI disks from non-IBM vendors to an RT running AIX 2.2.1. 1) Get the geometry information for the drive and create a ddi stanza file with the proper information. A) Use as a reference the ddi information for IBM's 9332 drive. Copy the /etc/ddi/9332 entry to one of your own name, such as /etc/ddi/wren7. NOTE: The commented lines in the ddi stanza are IMPORTANT. do not change them, or remove them. They must follow AFTER the entry being described. Find the following entries and change them to match the characteristics of your SCSI drive: > If the the drive is confused about syncronous operations set this to 'true' and the driver will not offer SYNC as an option while setting up the data transfers: (No Negotiate For Syncronous Transfer) == nnfst nnfst = false > If necssary, adjust the number of allowed bad blocks. noabb = 600 * Number of Allowed Bad Blocks > Here is one of the CRITICAL parameters! Make sure that this value matches the drives actual number of usable blocks on the device, *LESS* any CE cylinders and reserved track areas. nobod = 391182 * Number Of Blocks On Device. > AIX filesystems and drives use an underlying physical block size of 512 bytes. All the drives I have used have been able to select 512 as the block size. (This value is specified in 256 byte increments in the ddi.. 2*256=512) nobub = 2 * Num of 256-Byte Units/Block > The drives actual number of sectors per track is used by AIX and VRM for constructing good accessing inode superblock structures, so set this value as per the vendors specifications. nospt = 17 * Number of Sectors per Track B) Change the string /d9332/ to /wren7/. I never went over 5 characters, just made sure that the strings started with a letter, and were 5 characters long. The value used for the string is not important, just be unique among the entries if you have more then one type of SCSI disk, and it matches the *NAME* of the file in /etc/ddi directory. C) Next, we have to setup some things by hand that the 'devices' command would normally do for us. The disk drives scsi address and the dma channel to use when accesing the SCSI card/device. * Find the stanza entries in your newly cloned ddi entry that now read: wren7scsi1: biopa = d50 cn = 6 * DMA Channel Number * 0, 1, 3, 5, 6, 7 wren7scsi2: biopa = 950 cn =5 * DMA Channel Number * 0, 1, 3, 5, 6, 7 - The entries that start 'wren7scsi1' sets the addressing parameters for accessing the SCSI card 1. If this is your very first attached SCSI device, make sure that the DMA channel number used is not in use by any other cards in your system. The 'biopa' and 'cn' paramemters MUST track togther. The 'cn' paramenter sets the dma channel number. The 'biopa' sets the base io address for the card. - Find the entry that reads: wren7scsi### * The following stanzas are 9332 DASDs on the first SCSI adapter wren7scsi100: biopa = d50 cn = 6 * DMA Channel Number * 0, 1, 3, 5, 6, 7 The name of the stanza is the entries to be used for the individual instance of a device. The name is wren7scsi%$# Where % = The card number (1 or 2) $ = The device SCSI ID (0-6) % = the device LUN ID (0-1) Make the same change to the 'cn' and 'biopa' values. (if necessary) D) Now comes the critical step, editing the /etc/system file. Make a bakup copy of /etc/system.. syntax errors in this file will make your system unbootable! Make sure that you have a working copy of your 'Installation/maint' diskette around to be able to get back in and repair any botched attempts. Add the 'hdisk' by manually adding the stanza(s) for the disk. This would normally have been done by the devices command. External disks start with 'hdisk6'. hdisk0,1,2 are internal on the first EESDI adapter. hdisk3,4,5 are on the possible portalble EESDI adapter. Pick a uunique IODN number for the stanza. I usually start with 2310. The scsi id and scsi card are pointed to by the 'nname=' and 'use=' stanzas of this entry. Make sure the 'file=', 'kaf_file=' and 'kaf_use=' stanzas point at your newly added 'ddi' file in /etc/ddi. hdisk6: * MAXSTOR XT-8760 SCSI Disk Drive aflag = true * IBM SCSI Adapter, Primary adp = scsi1, scsi2 config = scsihelp crname = false dname = hdisk driver = vscsi dtype = scsi * IBM RT PC SCSI Devices file = /etc/ddi/max iodn = 2310 kaf_file = /etc/ddi/max kaf_use = kmax maxdev = 5 name = max noddi = false nodl = false noduplicate = false noipl = true noshow = false nospecial = true shared = true specproc = /etc/scsiproc switchable = false vint = 4 modes = rw-rw---- nname = maxscsi130 use = dmaxscsi130 E) Activate the usage of the new SCSI disk by the 'varyon' command. The first time you issue this command, the varyon may fail on some SCSI disks. The pending Unit Check from coming ready will confuse the vary commands processing. The vary processing will discover (hopefully) that the minidisk partition table is missing, and will proceed to create one. Depending on the formatting or prior data on the drive, this step can fail, if the varyon command gets confused about if a valid minidisk partition table is present. If this happens, take the drive to some KIND soul and have them write ZEROS on the first 100 sectors of the drive, and re-attach, and re-issue a varyon command (most likely twice). If everything is correct the vary will complete without errors, and a subsequent 'minidisks' command will show that space as available. F) Modify the system to varyon the SCSI disk(s) on each boot. Add the following command to the end of /etc/rc.include varyon -q -d hdisk6 varyon -q -d hdisk7 (or however many disks you have) You might have to issue the command twice for each drive. [ Note: It has been a while since I have personally done this, but the general procedure does work. I will re-verify it shortly. - MW ] ------------------------------ Q.15 - The diagnostics start booting ok, then fail with a LED code. What do the codes mean? This problem can caused by several problems among which are: 1) Unrecognized floppy controller. 2) Unrecognized or missing RT approved display adapter. 3) Serious malfunction one or more adapters. 4) Serious malfunction in a portion of the processor or memory cards installed. Usually, the failing component code will be the last displayed LED value. Reference the Hardware FAQ for a list of POST LED codes. These additonal LED codes can be found for problem areas: Steady code LED values: 22 - Boot diskette not found or failing. 25 - Software problem during soft IPL, power off and retry. 81 - 5085 failure, make sure it is powered on, and cables secure. 94 - Failing/missing display adapter. Use OEM display adapter diskette to continue. 99 - Keylock is in locked position. Blinking code LED values: 02 - Damaged diagnostic diskette, try another. 30 - Damaged diagnostic diskette, try another. 99 - Undetermined Problem, Possible failing: Keyboard, Diskette error, FP unit, memory card, or processor card. ------------------------------ Q.16 - How do I setup and use the AT Simulator. [ Open for adoption by someone very familiar with PC Simulator experience! - MW ] =============================================================================== *** FAQ Continued in AIX Question/Answer Part 2 *** ===================== end of ibm-rt-faq/aix-v2/misc/part1 ===================== -- Posted around the first of each month to comp.sys.ibm.pc.rt, and to comp.unix.aix. This FAQ is dedicated to use of the AIX Version 2 operating system, running on the IBM RISC processor known as the RT. The RT is known also as an IBM model 6150 or 6151 machine. For AIX RT submissions E-mail to: aix-rt@antimatr.hou.tx.us For AOS RT submissions E-mail to: aos-rt@antimatr.hou.tx.us For MACH RT submissions E-mail to: mach-rt@antimatr.hou.tx.us For general Q&A submissions (hardware and other stuff) E-mail to: faq-rt@antimatr.hou.tx.us Other comments and correctons should be sent to the author: Mark Whetzel DOMAIN addr: markw@antimatr.hou.tx.us BANG address: ..!uhnix1!lobster!antimatr!markw Daytime work number 8AM-5PM M-F CST VOICE: (713) 963-2544 -- AIX..... NOT just another UNIX. (tm) Mark Whetzel | My own RT system.. My own thoughts.. DOMAIN: markw@antimatr.hou.tx.us | IBM RT/135 running AIX 2.2.1 UUCP ..!menudo!lobster!antimatr!markw | comp.sys.ibm.pc.rt FAQ maintainer.