ULTRIX ROOT DISK RECOVERY PROCEDURE USING DECNSR WARNING: This procedure assumes the user is UNIX knowledgeable and has experience installing ULTRIX. FILES: (found on appropriate nsr server, mon[1-3].[tbu,rol]) /nsr/info/client-partition-tables. NOTES: Lines beginning with "//" are comment/descriptionof a step in the procedure. Lines beginning with ">", ">>", or "#" are system prompts, user should enter the text provided (with some interpretation). EXAMPLE TYPES USED: The failed disk is on adm1.tbu.mcirbn.com. The nsr server is mon1.tbu.mcirbn.com. The system disk is rz2. -------------------------------------------------------------------------------- // On nsr client (the system that lost its root (SYSTEM) disk) (assumed to be // halted). A CD drive must be connected to the client, the correct ULTRIX CD // version loaded into the drive. // If the unit has a graphical monitor/head, you must hook up a console cable if // it is to be done remotely. A 'telnet tsvx.xxx 200x' must be done to connect. >> printenv // record boot value on paper >> printenv [3,5] // record devices numbers and types, use either 3 or 5 >> setenv haltaction h // enable halt on power-on or init >> boot 3/rz4/vmunix // boot the network or the CD (default CD is rz4) // Select 3) System management (answer 3) // The following few commands will re-establish the original partition table on // the new root disk. The values are found on nsr server in // /nsr/info/client-partition-tables. # cd dev # ./MAKEDEV rz2 # /etc/chpt -q rrz2a # /etc/chpt -v -pa 0 50782 -pb 50782 379843 -pe 430625 195313 -pf 625938 390625\ -pg 1016563 839844 -ph 1856407 194453 rrz2a # ^D // select 2) ADVANCED Installation (answer 2) // select rz2 for root filesystem (answer 1 and y) // select 2) Existing partition table (answer 2 and y) >> boot 3/rz2/vmunix -a // enter system name (Ex. answer NODENAME (its short name) and y) // enter date/time (Ex answer 9705131503) // Select Eastern (answer e) // select Daylight Savings time (answer y, the default) // select USA (answer u, the default) // enter superuser password // CONSULT the file /nsr/info/client-partition-tables to determine // how to answer the following file system questions. The file will // have which drive and partition to use for each of the file systems. // the following answers are an example of one possible combination. // select rz2 for /usr filesystem (answer 1 and y) // select partition g (answer g) // select rz2 for swap1 space (answer 1 and y) // select partition b (answer b) // check the partition table or similar machine to see if seconday swap is used // select rz2 for crash dump space (answer 1 and y) // select partition b (answer b) // select rz2 for /var filesystem (answer 1 and y) // select partition h (answer h) // select 2) Color display (answer 2) // answer yes to UWS (Ultrix Workstation Software) // for subset selection, select 'Mandatory Only' // agree to install the Generic Kernel (answer y) // for CONFIGURATION FILE KERNEL OPTION SELECTION // select 'None of the above' // you do NOT want to edit configuration file (answer n, the default) // the system will attempt reboot and fail back to console ------------------------------------------------------------------------------- // Continuing on the client, for clients (i.e., ADMs, MONs, and NMSs) that have // graphics hardware, use this section to allow the new ULTRIX to use the // alternate console port (because ULTRIX defaults to the graphics screen). // bootup to single-user mode, DON'T append "-a" >> boot 3/rz2/vmunix # ed /etc/ttys /console .s/off/on/p :$ .s/:0/#:0/p w q # ^D # enter multi-user mode ------------------------------------------------------------------------------- // Continuing on the client, login as root, consult the partition table file to // determine what other partitions on the replaced drive must be initialized. // Here we are doing /usr/mcc on partition /dev/rz2f # newfs /dev/rrz2f rz26 # mkdir /usr/mcc # mount /dev/rz2f /usr/mcc // NOTE!! Make sure you mount all partitions you intend to restore via nsr. // If a partition is NOT mounted and you do a recover, you will fill up the // partition that it has it mount point in (usually root (/)). # mount /dev/rzxx /xxx // Look at /etc/ifconfig.stub from a similar node. Run the commands // substituting the // IP and HOSTNAME with correct information. It does // not have BIND so use IP in place of the name. // # /etc/ifconfig ln0 adm1-e4.rol.mcirbn.com netmask 255.255.255.0 // # /etc/ifconfig ln1 adm1-e5.rol.mcirbn.com netmask 255.255.255.0 // // to (where 111.111.111.111 is the actual IP address) // // # /etc/ifconfig ln0 111.111.111.111 netmask 255.255.255.0 // # /etc/ifconfig ln1 111.111.112.111 netmask 255.255.255.0 # /bin/hostname adm1.rol.mcirbn.com # /etc/ifconfig ln0 111.111.111.111 netmask 255.255.255.0 # /etc/ifconfig ln1 111.111.112.111 netmask 255.255.255.0 # /etc/ifconfig lo0 localhost // Start inetd # inetd // Put appropriate lines the /etc/hosts. Needs its own entry and the nsrhost. // You will appending the lines in the following examples, typing them in as // appropritate. You may use vi instead. # cat >> /etc/hosts > 111.111.111.111 adm1.tbu.mcirbn.com > 111.111.111.111 mon1.tbu.mcirbn.com nsrhost > ^D # // add nsr server to /.rhosts # cat >> /.rhosts > mon1.tbu.mcirbn.com > ^D # // if not in the datacenter, add a default route // where address is a bbr in the POP #route -n add default 167.31.141.11 1 ------------------------------------------------------------------------------- // Continuing on the DECnsr server, install the DECnsr Client license. Then // download the DECnsr client software to the client. # rcp /nsr/info/decnsr-net-clnt.pak adm1.tbu:/tmp # rsh adm1.tbu /tmp/decnsr-net-clnt.pak # setld -c BRXCDECRISC211 XINSTALL // select 1) To client disk (answer 1 and y) ... Client hostname: adm1.tbu ... Enter executable directory [ /usr/bin ]: ... Client hostname: ... Do you wish to continue? (y/n) [ y ]: ------------------------------------------------------------------------------- // Continuing on the client, do a rehash so that the new DECnsr command can be // found, then run the recover command to restore the original file systems. # rehash // Make room on root partition # rm /genvmunix // -s SERVER name (use full name since BIND is not running) // -f force overwrites of any existing file // -t time frame you want to start from, if you don't specify a time frame, // it will get the most current file version for each file # recover -s mon2.rol.mcirbn.com -f -t 2-may-1997 // add the file systems that originally existed on the drive (ADM) DECnsr> add / /usr /var /usr/mcc (INP) DECnsr> add / /usr /var /usr/mcc // Recover filesystems. Recover will complain that there isn't enough room, // agree to proceed (answer y) DECnsr> recover // lots of output DECnsr> quit // NOTE: If being done remotely and if unit has graphics head, then upon next // bootup, you will not have the control as it will go back to the graphics // head (/etc/tty). # shutdown -h now >> boot // Remove console cable (if installed for the recover) ------------------------------------------------------------------------------- // now do any updates (i.e., entitlement files) to this node that occurred // since the last backup (which was probably the previous night). // Done.