Some more random solaris/unix/geek stuff

Latest OS update: http://www.sun.com/software/solaris/get.jsp

Sun contract info:
go to http://ibis.sun.com/support and login
click on Account
Do a blank search


NetBackup client install (need win32 base and to untar packages)
first install base
then install the updates

ILOM
serial connect
cd /SP/network
then set pending properties to how you want them, then set commitpending
= true and it'll set the properties
set pendingipdiscovery=static
set other "pending" properties
then cd to other SP services and enable the ones you want
set ilom password

backup speed
put 65536 in /usr/openv/netbackup/net_buffer_sz
adjust /usr/openv/netbackup/SIZE_DATA_BUFFER and NUMBER_DATA_BUFFERS

package updater for RHEL 5.4 (Red hat enterprise linux) is pup. Seems to work pretty well.

Solaris 10 restore

Here's one that surprised me quite a bit. I had a server that was toast and had to burn it down and rebuild it. The first restore I did wasn't set up right, but I didn't know that at the time. When the server eventually came up, it seemed to work just fine. It's an all in one server so it does apache, coldfusion, mysql, dns, sendmail, and some other apps. All of those appeared to be functioning correctly, so I went on to work on other stuff that was switched as part of the disaster recovery plan. Then I went to check something on the production server and found that in the local zone running the webserver and most of the rest of the stuff, I was missing some pretty seriously important and low level commands from the OS. Things like vi, touch, uptime, and a bunch of others. Baffling. Most baffling for me was how it was running at all with so many low level OS bits missing.

Turns out with netbackup I needed to check "overwrite files" and "rename hard links" but not the softlinks. The hard links hadn't been checked the first time through, so it didn't restore any of the hard link files. Or at least, thats the nearest explanation I can come up with, because checking that and redoing the restore (which was a nightmare in its own right that I'll write up later) seems to have fixed everything just fine.

big bunch of random stuff

cold fusion has a file named registry that you can set the admin password to blank and then reset the admin passwd.

---------------------------------------------------
When cloning one sun machine to another with different hardware:
boot from cd/dvd and partition the drives correctly
newfs the drives
ufsrestore from the tape
ufsdump the new drives
modify /etc/hosts /etc/nodename /etc/hostname.<eth#> for new info
ok boot cdrom -s
...
# mount /dev/dsk/c0t3d0s0 /a <--- t3 is being used as an example only
# cd /tmp/dev
# tar cvfp - . | ( cd /a/dev; tar xvfp - )
# cd /tmp/devices
# tar cvfp - . | ( cd /a/devices; tar xvfp - )
# cd /tmp/root/etc
# cp path_to_inst /a/etc/path_to_inst

NOTE: The vfstab file may need to be edited at this time to make
any
changes to the target address

# cd /tmp/root; umount /a; halt
...
ok boot -rv

------------------------------------------------------------------
find cd rom on solaris
cfgadm -al | grep CD
then as an example of vfstab:
/dev/dsk/c1t6d0s0 - /cdrom hsfs - no -

--------------------------------------------

setting up mirrors on v210s
format drives to make partitions the same
metadb -a -c3 -f c1t0d0s7
metadb -a -c3 c1t1d0s7
vi /etc/lvm/md.tab and list mirrors
metainit -f d11
metainit -f d21
metainit -f d31
metainit -f d41
metainit d10
metainit d20
metainit d30
metainit d40
edit vfstab to boot to meta devices instead of slices and preserve / slice vfstab entry
metaroot d10
init 6
metainit d12
metainit d22
metainit d32
metainit d42
metattach d10 d12
metattach d20 d22
metattach d30 d32
metattach d40 d42
metastat d10
swap -l
dumpadm
dumpadm -d /dev/md/dsk/d20
metastat | grep -v "State: Okay"

If you have to boot back to the slice instead of the metadevice remember that /etc/system has a line about the root boot device or use "metaroot /dev/dsk/c1t0d0s0"

---------------------------------------------------------
jobs is the command to see fg/bg jobs
smpatch analyze to get a list of patchs available.
showrev -p then grep for patch number to see current patchs
updatemanager is the patch manager on solaris 10
<cfgadm -c [un]configure c1> configs or unconfigs device list c1 shown by
command <cfgadm -al>
devfsadm finds new hardware attached
iostat -En displays all device errors in descriptive format
-----------------------------------------
netbackup commands:
bpps -a shows all running processes
jnbSA is the graphical interface to netbackup
bp.kill_all to kill all processes (not elegant)
get_license_key to add/delete/list licenses
sgscan to scan for devices

---------------------------------------

When mysql gets the error:
"Database page corruption on disk or a failed
InnoDB: file read of page"
delete the ibdata1, ib_logfile* and then restart mysql and it should work.
There may also be:
"Warning: an inconsistent page in the doublewrite buffer" errors.

----------------------

mrtg for throughput thru switchs

robtest - netbackup robot test software

--------------

ma - to mark a spot in vi
d'a - deletes back to that spot