Unter http://www.sunfreeware.com findet man "ready to use"
Software für SUN Solaris, wie beispielsweise TOP, AMANDA, GCC, GDB etc. Download via
FTP von: «ftp://nce.sun.ch/pub/freeware/sparc/7»
CD-ROM Solaris Intranet Extension (siehe auch Solaris Server Intranet
$ cd /cdrom/cdrom0/nfsc/sparc
$ pkgadd -d `pwd`
$ uname -u
$ ifconfig -a
$ eject cdrom
$ prtvtoc /dev/rdsk/....
/dev/rmt/0l (tar cvf /dev/rmt/0l)
/etc/system definieren konfigurieren (Prestoserve, Oracle,
Logfile der Installation: /var/sadm/system/logs/install_log
Installation über serielles Terminal an Nullmodem Kabel
OK boot cdrom - w (Terminal an ttya)
$ init 0
OK boot -s
$ fsck -m /dev/dsk/c0t0d0s0 (und übrige Filesysteme)
$ tar cvf /dev/rmt/0l
Anzeige der installierten Packages
Check ob Package SUNWpcnfd correct installiert ist
$ pkgchk -v SUNWpcnfd
Package installieren (Path ist meistens /cdrom/cdrom0/....)
$ pkgadd -d <Path to CDROM> SUNWpcnfd
Das Package SUNWpcnfd entfernen
$ pkgrm SUNWpcnfd
In single user mode booten
OK boot -s
Kernel zwingen /devices neu aufzubauen nach dem Anschluss von neuer
OK boot -r
OK boot -v
Vom CD-ROM aus booten: Notboot !
OK boot cdrom
Angeschlossene SCSI-Geräte testen
List System Devices, e.g. SUNW,hme = Sun Fast Ethernet PCI Adapter
List Network Devices
Monitoring Network Activity
OK apply watch-net <full path name of the hme interface, see
Monitor Variablen ändern, anzeigen
OK eeprom ttya-mode=38400,8,n,1,h
Welche Kernel-Module sind geladen ?
Logfile von syslog
Konfiguration des syslog Daemon
Directory mit Default files
Remote root logins erlauben
Master-File beim Booten
Run-Level Start/Stop Files
Scripts für Run-Levels
$ shutdown -g0 -i0
$ reboot (entspricht init 6)
Konfiguration der LAN-Interfaces
$ ifconfig -a
Netmask setzen: siehe /etc/netmasks
Jedes LAN-Interface hat /etc/hostname.le0 mit Hostnamen
LAN-Setup: /etc/rcS.d/S30rootusr.sh (Interfaces konfigurieren)
/etc/rc2.d/S72inetsvc (LAN konfigurieren)
Phys Addressen nachschauen
$ arp -a
Net to Media Table
Flags Phys Addr
------ -------------------- --------------- ----- ---------------
255.255.255.255 SP 08:00:20:89:27:03
Alle hosts im Netz 184.108.40.206 werden über den Router
$ route add net 220.127.116.11 18.104.22.168 1
Der host 22.214.171.124 wird über den Router 126.96.36.199
erreicht. Siehe File /etc/rc2.d/S79staticroutes.
$ route add host 188.8.131.52 184.108.40.206 1
Routing Tabelle kontrollieren
$ netstat -nr
Angabe des DNS Nameservers
Siehe Solris2 Administration Seite 103 und ff
/etc/vfstab (Soll) --> /etc/mnttab (Ist)
mount -F nfs -o bg,ro,soft gondwana:/usr/software /software
NFS-Server wird in /etc/init.d/nfs.client start gestartet.
Anzeige welche Directories gondwana zum mounten freigegeben hat
RESOURCE SERVER ACCESS TRANSPORT
gondwana:/export/home/zahn gondwana - -
gondwana:/export/home/steiner gondwana - -
/etc/dfs/dfstab (Soll) --> /etc/dfs/sharetab
Directory read-only freigeben
$ share -o ro /usr/software
Alle Directories in /etc/dfs/dfstab freigeben
Alle Directories in /etc/dfs/dfstab zurücknehmen
NFS-Server wird gestartet in
Anzeige der freigegbenen lokalen Direcories
Anzeige welche Clients nutzen welche Directories eines NFS-Servers
$ dfmounts -F nfs gondwana
RESOURCE SERVER PATHNAME CLIENTS
gondwana /export/home/zahn paragon.glue.ch,rabbit.glue.ch
- /etc/auto_master (Master Map konfigurieren)
- /etc/auto_home (Home Direcories verwalten)
- autofs ist ein spezielles Filesystem
- automount -v (Nach einer Aenderung an einer Map ausführen)
Siehe spezielles Dokument
Beispiel: SCSI-Disk Seagate ST150176L, 50MB an SUN Ultra Enterprise
disk_type = "Seagate ST150176L" \
: ctlr = "SCSI" \
: ncyl = 12022 : acyl = 2 : pcyl = 12024 : nhead = 22 : nsect = 369 \
: rpm = 7200 : bpt = 188928
* SCSI-Disc Konfiguration
Disk anschliessen, SCSI-Adresse kontrollieren, Terminierung
Unbedingt kontrolieren, dass eine SCSI-Adresse nicht mehrfach belegt ist. Dazu kann
meistens hinten am Gerät ein Tippschalter eingestellt werden. Man beacht, dass
in der Regel das letzte Gerät terminiert werden muss.
Disk formatieren (nur wenn notwendig !)
In der Regel muss eine Disk nicht neu formatiert werden, ist dies jedoch notwendig so
steht unter Solaris das Utility format zur Verfügung.
AVAILABLE DISK SELECTIONS:
0. c0t0d0 <SUN2.1G cyl 2733 alt 2 hd 19 sec 80>
1. c0t1d0 <SUN2.1G cyl 2733 alt 2 hd 19 sec 80>
2. c0t2d0 <IBM-DDRS-39130-S71D cyl 8186 alt 2 hd 10 sec 218>
3. c0t4d0 <SEAGATE-ST118273N-5764 cyl 7499 alt 2 hd 20 sec 237>
4. c0t5d0 <SEAGATE-ST150176LW-0002 cyl 12022 alt 2 hd 22 sec 369>
Specify disk (enter its number): 4
AVAILABLE DRIVE TYPES:
0. Auto configure
2. Seagate ST118273
3. Seagate ST150176L
4. Quantum ProDrive 80S
5. Quantum ProDrive 105S
6. CDC Wren IV 94171-344
Specify disk type (enter its number): 21
format> format (confirm with "yes")
Dadurch wird die Disk in logische Teile unterteilt. Jeder teil
enthält ein eigenes Filesystem.
Nun die Partitionierungsdaten eingeben, zB
Current partition table (original):
Total disk cylinders available: 2733 + 2 (reserved cylinders)
Part Tag Flag Cylinders Size Blocks
0 root wm 0 - 204 152.15MB (205/0/0) 311600
1 swap wu 205 - 377 128.40MB (173/0/0) 262960
2 backup wm 0 - 2732 1.98GB (2733/0/0) 4154160
3 home wm 378 - 1017 475.00MB (640/0/0) 972800
4 unassigned wm 0 0 (0/0/0) 0
5 unassigned wm 1018 - 1928 676.13MB (911/0/0) 1384720
6 usr wm 1929 - 2732 596.72MB (804/0/0) 1222080
7 unassigned wm 0 0 (0/0/0) 0
newfs -v -m 0 /dev/rdsk/c0t5d0s0
Damit wird ein Filesystem mit 0 % Min-Free auf der Partition 0 der
Disk an der SCSI-Adresse 5 erstellt.
Dazu den folgenden Eintrag in /etc/vfstab vornehmen
fsck mount mount
#to mount to fsck
pass at boot options
/dev/dsk/c0t5d0s0 /dev/rdsk/c0t5d0s0 /u02 ufs 6
Multiply the Blocks column by 512
$ swap -l
dev swaplo blocks
/dev/dsk/c0t0d0s1 32,1 16 262944 262944
262944 * 512 = 134 MB
$ showrev -p
Note, that Patchlevel 105181-15 is minimal needed for Oracle 8.1.6
- Download the Patch from: http://sunsolve.sun.com
- Read the README File included in the Patch
- Usually the only thing you have to do is:
$ cd <patch cluster directory>
$ cat /var/sadm/install_data/<luster
$ showrev -p
Reboot the system
You can trace system calls with truss on Solaris an strace on Linux
$ truss svrmgrl
If you suspect troubles with your Solaris device files, e.g. system
doesn't boot after a filesystem check, you may repair the solaris system using the
Halt the system immediately with the keys STOP-A, you will now see the
boot prompt: OK
Reset the machine with
Boot the machine with
OK boot -r
The command boot -r will rebuild all devices files according to your
attached hardware. If you cannot boot the machine, you can try the following commands:
drvconfig, disks, tapes
drvconfig - configure the /devices directory
The default operation of drvconfig is to create the /devices directory
tree that describes, in the filesystem namespace, the hardware layout of a particular
machine. Hardware devices present on the machine and powered on as well as pseudo-drivers
are represented under /devices. Normally this command is run automatically after a new
driver has been installed (with add_drv(1M)) and the system has been rebooted.
disks - creates /dev entries for hard disks attached to the
Disks creates symbolic links in the /dev/dsk and /dev/rdsk directories
pointing to the actual disk device special files under the /devices directory tree.
tapes - creates /dev entries for tape drives attached to the
Tapes creates symbolic links in the /dev/rmt directory to the actual tape device special
files under the /devices directory tree. Tapes searches the kernel device tree to see
what tape devices are attached to the system.
Here are some short tips for common tasks on SUN Solaris
2.6, 7 and 8
-r # Show Run
$ /usr/sbin/prtconf # Print the complete system configuration
$ /sbin/mountall -l # Mount all local filesystems.
$ /sbin/init S # Changing to single user mode
# /etc/mnttab: Contains information about devices
# are currently mounted. If there are mounted filesystems
# with quotas enabled, display them
if /usr/bin/cut -f 4 /etc/mnttab | \
/usr/bin/egrep '^quota|,quota' >/dev/null 2>&1; then
echo 'There are mounted filesystems with quotas enabled'
# You will also need to uncomment the sa entries in
# the system crontab /var/spool/cron/crontabs/sys.
# Refer to the sar(1) and sadc(1m) man pages
# for more information.
$ /usr/bin/su sys -c "/usr/lib/sa/sadc
# sysidtool is a suite of five programs that configure a
# system, or one that has been unconfigured with sys-
# unconfig(1M). The sysidtool programs run automatically at
# system installation, or during the first boot after a
# machine has been successfully unconfigured.
# These programs have no effect except at such times, and
# should never be run manually.
# System Files are
Configure /etc/asppp.cf for the aspppd daemon
$ /usr/sbin/aspppd -d 1
# Getting Parameters Supported By The TCP Driver
# To see which parameters are supported by the TCP driver,
# use the following command:
$ ndd /dev/tcp \?
# The following command sets the value of the parameter
# ip_forwarding in the IP driver to zero. This disables IP
# packet forwarding.
$ /usr/sbin/ndd -set /dev/ip ip_forwarding 0
$ /usr/sbin/ndd -set /dev/ip ip_forwarding 1
# Configure default routers using the local
# configuration file. The file can contain the hostnames or IP
# addresses of one or more default routers.
# The default routes listed in the "/etc/defaultrouter" file will
# replace those added by the kernel during diskless booting. An
# empty "/etc/defaultrouter" file will cause the default route
# added by the kernel to be deleted.
# Note that the default router file is ignored if we received routes
# from a DHCP server. Our policy is to always trust DHCP over local
# Set Default Route
$ route -n add default <Default Route from /etc/defaultrouter>
# Show Default Route
$ /usr/sbin/route -fn
default 220.127.116.11 done
if [ -f /etc/defaultdomain ]; then
/usr/bin/domainname `cat /etc/defaultdomain`
echo "NIS domainname is `/usr/bin/domainname`"
# rpcbind - universal addresses to RPC program
# rpcinfo - report RPC information
# keyserv is a daemon that is used for storing the
# encryption keys of each user logged into the system. These
# encryption keys are used for accessing secure network ser-
# vices such as secure NFS and NIS+.
# If this machine is configured to be an
# Domain Name System (DNS) server, run the name daemon.
# Start named prior to: route add net host,
# to avoid dns gethostbyname timout delay for
# nameserver during boot.
if [ -f /usr/sbin/in.named -a -f /etc/named.conf ]; then
echo 'starting internet domain name server.'
Configuration File: /etc/syslog.conf
Message File: /var/adm/messages
# How to setup IP-Alias on SUN
1. Setup File /etc/hostname.hme0:1 for the second IP-Address
2. Insert IP-Address in /etc/hosts
# Internet host table
ux-portal1 # IP-address on hme0:0
ldap # IP-alias
3. Start alias IP-Address on Interface in /etc/rc2.d
S99ipalias -> ../init.d/ipalias
# Akadia AG, Arvenweg 4, CH-3604 Thun
# File: ipalias
# Autor: Martin Zahn / 10.05.2000
# Purpose: Setup second IP address on hme0:1
if [ -f /etc/hostname.hme0:1 ]
case "$1" in
'start') # Start second IP
address on hme0:1
echo "Start multi-homed server for UX-ALIAS1 on hme0:1"
ifconfig hme0:1 18.104.22.168 up
'stop') # Stop second
IP address on hme0:1
echo "Stop multi-homed server for UX-ALIAS1 on hme0:1"
ifconfig hme0:1 22.214.171.124 down
4. Check IP-Address on second Interface
The Solaris automount utility installs autofs mount points and associates an automount map with each mount
point. The autofs file system monitors attempts to access
directories within it and notifies the automountd daemon.
The daemon uses the map to locate a file system, which it then mounts at the point of
reference within the autofs file system. You can assign a
map to an autofs mount using an entry in the /etc/auto_master map or a direct map in /etc/auto_direct. If the file system is not accessed within an
appropriate interval (five minutes by default), the automountd daemon unmounts the file system.
The mount point /net is by
default the location, where automountd mounts NFS
filesystems, which are exported on other machines. Lets suppose, that you have the
filesystem /home exported on the NFS server saphir, then the (Solaris) NFS client with an active automounter
will automatically mount this NFS filesystem under /net/saphir/.
You probably doesn't want this default behavior. If you
insert the following entry in /etc/auto_direct ....
.... then, the directory /local on the remote machine "remote_machine" will be mounted on the
local machine under /opt/local.
The Solaris utility kbd manipulates the state of the keyboard or display the type
of keyboard or change the default keyboard abort sequence effect. Suppose, that
you do not want that everybody can halt the the system you must change the default value.
We also noticed, that the Solaris machines attached to a switch box, using a character
terminal on a serial line, may halt when you switch from one machine to the other.
kbd [ -r ] [ -t ] [ -c on|off ]
[ -a enable|disable ] [ -d keyboard device ]
kbd -i [ -d keyboard device ]
kbd manipulates the state
of the keyboard, or displays the keyboard type or allows the default keyboard abort
sequence effect to be changed. The default keyboard device being set is /dev/kbd.
The -i option reads and processes default values for the
keyclick and keyboard abort settings from the keyboard default file, /etc/default/kbd. Only keyboards that support a clicker respond to
the -c option. If you want to turn clicking on by default, add or change the current
value of the KEYCLICK variable to the value on in the keyboard default file, /etc/default/kbd, as shown here.
Then, run the command 'kbd
-i' to change the current setting. Valid settings for this variable are the
values on and off. Other values are ignored. If the variable is not specified in the
default file, the setting is unchanged.
The keyboard abort sequence (L1-A or STOP-A)
on the keyboard and BREAK on the serial console input device on most systems)
effect may only be changed by the superuser, using the
-a option. On most systems, the default effect of the keyboard abort sequence is to
suspend the operating system and enter the debugger or the monitor.
If you want to permanently change the software default
effect of the keyboard abort sequence, you can add or change the current value of the
KEYBOARD_ABORT variable to the
value disable in the keyboard default file, /etc/default/kbd, as shown here.
Then, run the command 'kbd
-i' to change the current setting. Valid settings for this value are the
values enable and disable. Other values are ignored. If the variable is not specified in
the default file, the setting is unchanged.
||Set keyboard defaults from the keyboard default file.
This option is mutually exclusive with all other options except for the -d keyboard
device option. This option instructs the keyboard command to read and process
keyclick and keyboard abort default values from the /etc/default/kbd file. This option can only be used by the
||Reset the keyboard as if power-up.
||Return the type of the keyboard being used.
||On/Off state Turn the clicking of the keyboard on or
||Enable/Disable state; Enable or disable the keyboard
abort sequence effect.
This chapter describes procedures for monitoring system
performance by using the vmstat, iostat, df, and
sar commands. This is a list of the step-by-step
instructions in this chapter.
The following example shows the vmstat
display of statistics gathered at five-second intervals.
$ vmstat 5
disk faults cpu
r b w swap free re mf pi po fr de sr f0 s3 -- --
in sy cs us sy id
0 0 8 28312 668 0 9 2 0
1 0 0 0 1 0 0 10 61 82
1 2 97
0 0 3 31940 248 0 10 20 0 26 0 27
0 4 0 0 53 189 191 6 6 88
0 0 3 32080 288 3 19 49 6 26 0 15
0 9 0 0 75 415 277 6 15 79
0 0 3 32080 256 0 26 20 6 21 0 12
1 6 0 0 163 110 138 1 3 96
0 1 3 32060 256 3 45 52 28 61 0 27 5
12 0 0 195 191 223 7 11 82
0 0 3 32056 260 0 1 0 0
0 0 0 0 0 0 0 4 52 84
0 1 99
Reports the following states:
The number of kernel threads in the dispatch queue
Blocked kernel threads waiting for resources
Swapped out LWPs waiting for processing resources to finish
Reports on usage of real and virtual memory:
Available swap space
Size of the free list
Reports on page faults and paging activity, in units per
Minor and major faults
Kbytes paged in
Kbytes paged out
Anticipated memory needed by recently swapped-in processes
Pages scanned by page daemon (not currently in use). If sr does
not equal zero, the page daemon has been running.
Reports the number of disk operations per second, showing data
on up to four disks
Reports the trap/interrupt rates (per second):
Interrupts per second
System calls per second
CPU context switch rate
Reports on the use of CPU time:
Run vmstat -s to show the total of
various system events that have taken place since the system was last booted.
0 swap outs
0 pages swapped in
0 pages swapped out
409376480 total address trans. faults taken
3075036 page ins
2601555 page outs
3812452 pages paged in
6525552 pages paged out
11007609 total reclaims
10927650 reclaims from free list
0 micro (hat) faults
409376480 minor (as) faults
2957386 major faults
102738273 copy-on-write faults
61711047 zero fill page faults
1002562077 pages examined by the clock daemon
7881 revolutions of the clock hand
16716370 pages freed by the clock daemon
741660225 cpu context switches
736047593 device interrupts
2496638575 system calls
430283487 total name lookups (cache hits 95%)
10484677 user cpu
9528364 system cpu
443762786 idle cpu
16281790 wait cpu
Run vmstat -S to show swapping
r b w swap free si so pi po fr de sr m1
m3 m4 m5 in sy cs us sy id
0 0 0 8512 888 0 0 12 21 55 0
417 1 0 0 0 206 1040 308 2 2 96
si = Average number of LWPs swapped in per second
so = Number of whole processes swapped out
You can display disk activity information by using the iostat command with a time interval. The following example
shows disk statistics gathered every five seconds.
tin tout kps tps serv kps tps serv kps tps serv kps tps
serv us sy wt id
0 2 10 1
28 2 0 22 0
0 0 1 0 10
2 2 3 92
0 47 58 7 39
16 2 34 0 0
0 0 0 0 0 2 19 78
0 16 0 0
0 0 0 0 0
0 0 0 0 0
0 1 0 98
0 16 0 0
0 0 0 0 0
0 0 0 0 0
0 0 1 99
0 16 2 0
22 0 0 0 0
0 0 0 0 0
2 3 1 95
0 24 0 0
0 0 0 0 0
0 0 0 0 0
0 1 1 98
For Each ...
Number of characters in the terminal input queue
Number of characters in the terminal output queue
Blocks per second
Transactions per second
Average service time, in milliseconds
In user mode
In system mode
Waiting for I/O
Run iostat -xtc to get extended
disk statistics. This command displays a line of output for each disk.
extended device statistics
device r/s w/s kr/s kw/s wait actv
svc_t %w %b tin tout us sy wt id
md1 0.4 0.9
3.6 6.9 0.0 0.0 27.7 1
1 0 2 2 2 3 92
md3 0.1 0.2
1.0 1.3 0.0 0.0 21.7 0
md4 0.0 0.0
0.0 0.0 0.0 0.0 0.0
md5 0.0 0.0
0.7 0.0 0.0 0.0 9.9
md8 0.8 0.3 6.7
14.2 0.0 0.0 13.1 0 1
md10 0.2 0.9 1.8
6.8 0.0 0.0 15.5 0 1
md11 0.2 0.9 1.8
6.8 0.0 0.0 14.8 0 1
md30 0.0 0.2 0.5
1.3 0.0 0.0 11.4 0 0
md31 0.0 0.2 0.5
1.3 0.0 0.0 10.2 0 0
md40 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0 0
md41 0.0 0.0 0.0
0.0 0.0 0.0 0.0 0 0
md50 0.0 0.0 0.4
0.0 0.0 0.0 9.4 0 0
md51 0.0 0.0 0.4
0.0 0.0 0.0 7.3 0 0
md80 0.4 0.3 3.3
14.2 0.0 0.0 10.3 0 0
md81 0.4 0.3 3.3
14.2 0.0 0.0 11.7 0 1
sd0 0.6 2.1 6.0
22.8 0.0 0.0 16.3 0 3
sd1 0.6 2.1 6.0
22.8 0.0 0.0 15.2 0 2
Reads per second
Writes per second
Kbytes read per second
Kbytes written per second
Average number of transactions waiting for service (queue
Average number of transactions actively being serviced
Average service time, in milliseconds
Percentage of time the queue is not empty
Percentage of time the disk is busy
Display CPU utilization with the sar -u command. (The sar command
without any options is equivalent to sar -u.) At any given moment, the processor is
either busy or idle. When busy, the processor is in either user or system mode. When
idle, the processor is either waiting for I/O completion or "sitting still" with no work
Measure CPU utilization during 5 secs one time.
sar -u 5 1
Measure CPU utilization during 60 secs 1440 times and write result in
sar -u -o sar.log 60 1440
To later review disk and tape activity from that period:
sar -d -f sar.log
Lists the percentage of time that the processor is in system
Lists the percentage of time that the processor is in user
Lists the percentage of time the processor is idle and waiting
for I/O completion
Lists the percentage of time the processor is idle and is not
waiting for I/O
A high %wio generally means a disk slowdown has occurred.
Solaris 7 and 8 include a native implementation of file
system journaling. This feature, known as "intent logging" or just "logging"
enables FASTER file system operation and FASTER system boot.
It's trivial to implement and safe to use. The new logging
feature is an option to the Unix File System (UFS), which is the standard file system for
all disk partitions on SUN servers, except for partitions holding swap space. By default,
the journaling option is disabled. Logging is enabled on a per file system basis, and it
can even be enabled on / (root file system) and other operating system partitions.
Solaris UFS logging works by allocating space from the
file system's free blocks. Within that space, all metadata changes to the file system are
written. Metadata includes directory and I-node information but not file data blocks,
essentially everything but the actual data within the file. So, for example, a "file
create" modifies the directory structure and allocates a new I-node, and those activities
are written to the logging space. Once the metadata changes are made to the logging area,
the system is free to perform other operations to the file system. In the background, the
information in the log is flushed to the file system and updates the appropriate
directory and I-node structures, completing the file system operations.
The logging data is written sequentially within the log
space. It's therefore much faster for the operating system to complete metadata changes
via logging and background flushing than by directly modifying the metadata (via random
I/O) spread across the disk. The size of the logging space is based on the size of the
file system, and equals 1 MB per 1 GB of file system space, up to 64 MB. The space is
used as a circular log: if the log space is about to fill up, new metadata change
requests are paused while the log is emptied. As changes are moved from the log to the
file system, that log space is made available, and new metadata changes can be written to
the logging space.
Usually with UFS, if the system crashes during any file
system operation, the entire system must have its consistency checked via the fsck
command. That command can take several minutes per file system because it checks all
metadata and file data to ensure the structures are correct, free, and used, and that the
I-node block counts are correct. It also confirms that the free space available is
current, repairs inconsistencies, and occasionally requires manual intervention to fix
large problems. Files and even directories can be lost, depending on the operations
occurring at the time of the crash.
Because metadata changes are made first to the log space
rather than to the file system, the consistency check for a logged file system after a
crash is a simple and fast operation. The system evaluates the logging data and
determines which changes had completed against the underlying file system, which had yet
to start, and which were in progress. Those completed or not yet started are removed from
the log, and those partly completed are either undone or completed. If there's sufficient
data in the log to complete the operation, it's completed. Otherwise, the changes made
are removed from the underlying file system.
People familiar with database operation will recognize the
similarity between database transaction processing and the activities here. The end
result is that the underlying file system is consistent, and no thorough consistency
checking is needed. That operation completes in a few seconds per file system.
Starting with Solaris 7, there's a new logging option to
the mount command and in the /etc/vfstab system configuration file. Logging only
appears in a couple other places within Solaris. The mount command shows which partitions
are mounted and lists logging in the options fields for each partition on which logging
is enabled. Finally, at system boot time, the fsck phase reports per partition whether
each is stable, logging, or being checked. There are no other status commands available
to determine the state of logging.
A = Device to mount
B = Device to fsck
C = Mount point
D = Filesystem Type
E = Fsck pass (unimportatnt with logging)
F = Mount at boot
G = Mount options
C D E
/dev/fd fd - no -
/proc proc - no -
- swap - no -
/dev/dsk/c0t0d0s0 /dev/rdsk/c0t0d0s0 /
ufs 1 no logging
/dev/dsk/c0t0d0s6 /dev/rdsk/c0t0d0s6 /usr ufs
2 no logging
/dev/dsk/c0t0d0s1 /dev/rdsk/c0t0d0s1 /var ufs
3 no logging
/dev/dsk/c0t0d0s7 /dev/rdsk/c0t0d0s7 /home ufs
4 yes logging
/dev/dsk/c0t0d0s5 /dev/rdsk/c0t0d0s5 /opt ufs
5 yes logging
/dev/dsk/c0t8d0s0 /dev/rdsk/c0t8d0s0 /u01 ufs
6 yes logging
/dev/dsk/c0t9d0s0 /dev/rdsk/c0t9d0s0 /u02 ufs
7 yes logging
/dev/dsk/c0t10d0s0 /dev/rdsk/c0t10d0s0 /u03 ufs
8 yes logging
/dev/dsk/c0t11d0s0 /dev/rdsk/c0t11d0s0 /u04 ufs
9 yes logging
/dev/dsk/c0t12d0s0 /dev/rdsk/c0t12d0s0 /u05 ufs 10
/dev/dsk/c1t13d0s0 /dev/rdsk/c1t13d0s0 /app ufs 11
/dev/dsk/c1t14d0s0 /dev/rdsk/c1t14d0s0 /users ufs 12 yes
/tmp tmpfs - yes -
Logging increases performance, decreases fsck time,
removes the risk of a file system corruption, can be used on all UFS partitions
(including root), and is free.
The log system messages daemon syslogd reads and forwards system
messages to the appropriate log files and/or users, depending upon the priority of a
message and the system facility from which it originates. The configuration file
/etc/syslog.conf controls where messages are forwarded. The syslogd daemon ignores
any faulty entry in /etc/syslog.conf, specially spaces instead of tabs are not recognized
by syslogd. Therefore always check the entries in /etc/syslog.conf in the debugging mode
# /etc/init.d/syslog stop
# /usr/sbin/syslogd -d
getnets() found 1 addresses, they are:
amiloghost() testing 126.96.36.199.2.2
syslogd: line 14: unknown priority name
syslogd: version 1.70
Started: Sat Jan 6 10:11:47 2001
Input message count: system 0, network 0
# Outputs: 10
5 3 3 3 5 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3
X CONSOLE: /dev/sysmsg
7 3 3 5 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 X FILE: /var/adm/messages
X X 6 X X X X X X X X X X X X X X X X X X X X X X UNUSED:
3 1 1 3 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 X USERS: operator
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 X USERS: root
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 X WALL:
X 3 X X X X X X X X X X X X X X X X X X X X X X X CONSOLE: /dev/sysmsg
X 3 X X X X X X X X X X X X X X X X X X X X X X X FILE: /var/adm/messages
X 1 X X X X X X X X X X X X X X X X X X X X X X X USERS: root, operator
X 0 X X X X X X X X X X X X X X X X X X X X X X X WALL:
Per File Statistics
Tot Dups Nofwd Errs
--- ---- ----- ----
0 0 0
off & running....
sys_poll blocking, init_cnt=0
# /etc/init.d/syslog start
Line 14 in /etc/syslog.conf are filled up with spaces instead of tabs.
Replace the spaces with tabs and syslogd will accept the new entry in Line 14.
If you check the oracle process with the OS comand "pmap" or
"top", you can see that each oracle process use more than 100M memory. Is this a
problem on the Oracle installation or something else? It seems that pmap counts the SGA
size as the private memory segment of each oracle process, but we believe the SGA size
should be shared.
Output from "top" on our Solaris System with Orcale 188.8.131.52
PID USERNAME THR PRI NICE SIZE
RES STATE TIME CPU COMMAND
361 oracle 258 59 0 124M 88M
sleep 0:01 0.00% oracle
373 oracle 11 59 0 122M 88M
sleep 41:50 0.00% oracle
363 oracle 11 59 0 119M 88M
sleep 0:01 0.00% oracle
365 oracle 11 58 0 119M 88M
sleep 0:17 0.02% oracle
359 oracle 1 59 0 119M
89M sleep 0:00 0.00% oracle
377 oracle 1 59 0 119M
88M sleep 0:00 0.00% oracle
375 oracle 1 58 0 119M
88M sleep 0:00 0.00% oracle
367 oracle 1 58 0 118M
89M sleep 0:00 0.00% oracle
371 oracle 1 58 0 118M
89M sleep 0:00 0.00% oracle
369 oracle 1 58 0 118M
88M sleep 0:00 0.00% oracle
On many UNIX platforms and specially on Sun platforms, the text of the
Oracle binary and shared libraries are actually shared between background processes if
these instances share the same ORACLE_HOME. So you need to subtract the shared text of
the oracle binary and the shared libraries in the result of the OS commands.
Even pmap and pmen utilities make mistakes between these memory divisions, and sometimes
SGA and text executable are often added incorrectly.
This can be used by anyone who has privleges for the pmap, which
can be found in /usr/proc/bin/. First, we need to find the process id (PID) of the
Oracle background process you wish to determine the memory size for. This is done
by issueing the following command:
# ps -u oracle -f
PPID C STIME TTY TIME CMD
oracle 359 1 0 12:26:17
? 0:00 ora_pmon_DIA3
oracle 361 1 0 12:26:17
? 0:01 ora_dbw0_DIA3
oracle 363 1 0 12:26:17
? 0:01 ora_lgwr_DIA3
oracle 365 1 0 12:26:17
? 0:18 ora_ckpt_DIA3
oracle 367 1 0 12:26:17
? 0:01 ora_smon_DIA3
oracle 369 1 0 12:26:17
? 0:00 ora_reco_DIA3
oracle 371 1 0 12:26:17
? 0:00 ora_snp0_DIA3
oracle 373 1 0 12:26:17
? 41:50 ora_s000_DIA3
oracle 375 1 0 12:26:17
? 0:00 ora_d000_DIA3
oracle 377 1 0 12:26:18
? 0:00 ora_d001_DIA3
Second, you then enter the following commands for the DB Writer process
(ora_dbw0_DIA3) with process id = 361 as an example.
# /usr/proc/bin/pmap 361 | grep "shmid"
80000000 82992K read/write/exec/shared [ shmid=0x2 ]
# /usr/proc/bin/pmap 361 | grep "total"
Then you take the total size: 124232K and subtract the SGA size which
the line marked with "shmid=" above, in this case it is 82992K. So, 124232K minus 82992K
is 41240K. So, the DBWR background process is approximately 41.2 MB. Repeat
this steps for all the background processes.
How much memory is needed on SUN Solaris? Explaining memory in Solaris
by reviewing the different types of memory and introducing a set of tools, the
Download the RMCmem tools available from ftp://playground.sun.com/pub/memtool. The
package includes a kernel module that provides extra instrumentation.
# cd /tmp
# zcat RMCmem3.8.2.tar.gz | tar xvf -
# pkgadd -d .
The package is installed in /opt/RMCmem (see README in this
Solaris is a virtual memory system. The total amount of memory that you
can use is increased by adding swap space to the system. If you ever see "out of memory"
messages, adding swap space is the usual fix. Performance of the system is very dependent
on how much physical memory (RAM) you have. If you don't have enough RAM to run your
workload, performance degrades rapidly.
Physical memory usage can be classified into four groups:
Kernel memory mapped into kernel address space
Process memory is mapped into a process address space
Filesystem cache memory that is not mapped into any address
Free memory that is not mapped into any address space
RMCmem includes a simple command to summarize this:
Kernel Memory: 60
Executable & libs: 42 Megabytes
Free, file cache: 11 Megabytes
The total physical memory can be seen using prtconf. Memory is
allocated in units called pages, and you can use the 'pagesize' command to see the size
in bytes per page:
# /usr/sbin/prtconf | grep Memory
Memory size: 1024 Megabytes
Kernel memory is allocated to hold the initial kernel code at boot
time, then grows dynamically as new device drivers and kernel modules are used. Kernel
tables also grow dynamically, unlike some older versions of Unix. As you add hardware and
processes to a system, the kernel will grow. In particular, to keep track of all the
memory in a system, the kernel allocates a page table structure.
If you have several gigabytes of RAM this table gets quite large. The
dynamic kernel memory allocator grabs memory in large "slabs," then allocates smaller
blocks more efficiently. This means that the kernel tends to grab a bit more memory than
it's really using. If there is a severe memory shortage, the kernel unloads unused kernel
modules and devices and frees unused slabs. The simplest summary of kernel memory usage
comes from sar. To show the kernel memory allocation (KMA) activities use (see man
sar for more details).
# sar -k 1
SunOS diamond 5.7 Generic_106541-12
sml_mem alloc fail
lg_mem alloc fail ovsz_alloc fail
6873088 6044236 0 44818432 43761720
0 11231232 0
Application processes consist of an address space divided into
segments, where each segment maps either to a file, anonymous memory (the swap space),
System V shared memory, or a memory mapped device. The mapped files include the code and
initialized data for the command and all its shared libraries.
What we really want to know, is the amount of RAM used by each segment.
This is shown by the pmem command in the RMCmem package.
# /opt/RMCmem/bin/pmem 361
Kbytes Resident Shared Private Permissions
82992 - read/write/exec [shmid=0x2]
16 8 8
16 8 8
8 8 -
.. . ............... ...........
80 72 8
8 8 -
16 8 8
8 8 -
-------- ------ ------ ------ ------
124232 93040 92728 312
Now we can see that the process address space size is 124232 kilobytes;
93040 kilobytes of that are currently resident in main memory, wherein 92728 kilobytes
are shared with other processes while 312 kilobytes are private. When this command
started only the 312 kilobytes of private memory were taken from the free list.
If we now go through all the processes on the system, add up how much
private memory they use, and also add in the shared memory for each mapped file, we'll
know how much application memory is in use. This summary is shown by prtmem as we
saw in the beginning, and the detail is listed by the memps command in RMCmem.
Resident Shared Private Process
... ....... ...... ......
359 118904k 93608k 92800k
367 118184k 93152k 92704k
369 117928k 93120k 92704k
371 118040k 93136k 92720k
365 119040k 93120k 92712k
377 118344k 93080k 92720k
363 119088k 93056k 92720k
375 118344k 93048k 92720k
361 124232k 93040k 92728k
373 121608k 93032k 92728k
This is the part of memory that is most confusing, as it is invisible.
You can only tell it's there if you access the same file twice and it is quicker the
The RMCmem package adds kernel instrumentation that counts up all the
pages for each cached file. The memps -m command lists the files that are
cached in order of the amount of memory they're consuming.
One problem is that within the kernel, the file is only known by its
inode number and filesystem mount point. The directory pathname for the file may not be
The RMCmem package tries to solve this problem by catching file names
as files are opened (by interposing on the vnode open code) and making an inode-to-name
lookup cache in the kernel. This cache size is limited (to 8192 entries by default), and
the file may have been opened before the kernel module was loaded, so it can't always
find the name.
# memps -m
Size InUse E/F
21064k 21064k F /usr
8184k 824k F /usr (inode 260922)
7752k 7752k F /usr (inode 540429)
7480k 7480k F /usr (inode 540428)
7480k 7480k F /usr (inode 540427)
6896k 6896k F /usr (inode 540450)
.... .... . .... ......
... and so on down to lots of files ...
# cd /usr
# find . -inum 540488
More infos about the RMCmem package can be found here as PDF
(Using the Doc online)
the Doc online)
Installation Instructions will cover the
installation of the DLT tape peripheral hardware and configuration of the system to
communicate with the DLT tape peripheral. In this example we use a «QUANTUM
The Solaris system must have the appropriate SCSI
interface for DLT drive to attached to, a SCSI single-ended DLT drive can be attached
only to a SCSI single-ended interface. The same is true for SCSI differential attachment.
Solaris includes a driver to efficiently communicate with SCSI tape drives, such as the
DLT tape peripheral.
Perform the installation as follows:
Shut down your Sun workstation/server and power off the machine and all scsi-devices.
Connect the DLT to the scsi-bus using good cables and make sure the bus is terminated
correctly. Set the scsi-id; id 4 or 5 are the most common to use.
If possible use a separate or underutilized SCSI bus for the DLT. Running the tape
drive on the same bus as the disk drives will never let you achieve any good throughput.
You bought the DLT because of performance didn't you ?
||Power on the devices/machine again and halt the boot
(or press the BREAK key if you have an ASCII
||Verify that the drive is connected properly.
Note: output from probe-scsi will not always be
correct if you enter the PROM monitor by breaking the boot process!
|Boot the system and log in as
root. When booting you should see a message similar to these: "st1: ".
Change directory to /kernel/drv.
Edit the st.conf file
by adding the following:
DLT7-data = 1,0x38,0,0x8639,4,0x82,0x83,0x84,0x85,3;
tape-config-list="<DLT tape unit>","<DLT reference name>","<DLTdata>"
is a variable defined by a series of tape configuration parameters listed below:
unit> is the vendor and product ID string for the DLT device.
Depending on the DLT tape peripheral you are installing, you must insert the
appropriate vendor and product ID for
<DLT tape unit> as described in the
DLT Tape Product
<DLT tape unit>
||QUANTUM DLT7000 (Total string
character count, including spaces, must equal 15).
name> is a name you select that the system will use to identify the DLT
device. This reference does not change the DLT product ID. When the system boots, the
reference name will be displayed in the list of peripheral devices recognized by the
<DLT-data> is a variable containing a series of additional DLT device configuration
information. You select a name in place of the <DLT
data> string. You will continue editing the st.conf file by defining the
name you selected for <DLT data>. The
definition depends on the DLT tape peripheral you are installing. For a
DLT7000 series unit add the following line:
data> contains 10 parameters and are described
|The first parameter, is the version
number and should not change.
The second parameter, designates the DLT tape type as defined
#define MT_ISOTHER 0x36 /* generic other type of tape drive */
#define MT_ISDLT 0x38 /* sun: SCSI DLT tape drive */
|The third parameter
is the block size. Since the DLT tape drive uses variable block
size, this value should be zero.
The fourth parameter, 0x8639, is a summation of values that
represent selected device options. The table below lists the options and the
st page has more
information about these and other possible device options. For certain
applications, it may be necessary to consider adding or removing one or more of
the device options.
|The fifth parameter, 4, defines the
number of densities. The maximum definable number of densities is 4.
|The sixth, seventh, eighth and
ninth parameter are used for system selection of tape densities. Use these
values for a DLT 7000 Tape Drive.
|The tenth parameter
defines which density the system will use as the default
density. The sixth, seventh, eighth and ninth parameters in the <DLT
data>string are referenced by the system as 0, 1, 2 and 3, respectively.
The 3 value for the tenth parameter selects the
0x85 density code as the system default density.
After editing the st.conf file, reboot the system:
|Reboot the System
The -r switch in the boot command enables a kernel compile and
includes the creation of device special files used for communication with the DLT
device. The -v switch enables verbose mode display of
system bootup. With verbose mode, the system should indicate that the DLT tape
peripheral is attached by displaying the <DLT reference name>string you
Enter the following command to verify the installation:
Product 'DLT7000 ' tape drive:
sense key(0x0)= No Additional Sense residual= 0
retries= 0 file no= 0 block no= 0
The target drive designations assigned by Solaris
may take on values higher than already established in the /dev/rmt/ path. This is not a problem but during a boot -rv, Solaris does not remove tape device files for drives that
are no longer attached to the system. This can increase the effort in locating the device
file for the configured drive, however, this can be minimized by first deleting the tape
then either boot the system with a:
or issue the following at the command line
drvconfig -i st; tapes
If the DLTtape is the only drive on the system, it's
target assignment should be zero. The Solaris man pages have more information on
drvconfig and tapes.
If you remove or add a device on Solaris then the devices files must be recreated,
either with boot -rv or devfsadm. For example to renumber the logical tape drive devices
do the following: Tape drives were numbered beginning with /dev/rmt/3 instead of /dev/rmt/0. The physical devices pointed to by the logical
/dev/rmt/ devices no longer existed, and we
wanted to renumber the valid devices beginning at /dev/rmt/0.
- Cleanup non-existent tape drive devices with devfsadm.
# devfsadm -C -c tape -v
- Remove all /dev/rmt logical links.
# rm -f /dev/rmt/*
- Recreate all /dev/rmt logical links with
# devfsadm -c
devfsadm(1M) maintains the /dev and /devices namespaces. It replaces the previous
suite of devfs administration tools including drvconfig(1M), disks(1M), tapes(1M),
ports(1M), audlinks(1M), and devlinks(1M).
The following options are supported:
|Cleanup mode. Prompt devfsadm to cleanup dangling /dev
links that are not normally removed. If the -c option is also used, devfsadm only
cleans up for the listed devices' classes.
|Restrict operations to devices of class device_class.
Solaris defines the following values for device_class: disk, tape, port,
audio, and pseudo. This option may be specified more than once to specify
multiple device classes.
The Solaris operating system gets the jumpstart for its booting from a hardware-level
interface called the OpenBoot PROM or OBP for short. OpenBoot at its heart has an
interactive command interpreter with a varied set of functions. OBP is a firmware which
is stored in the socketed startup PROM of the computer and consists of two parts, the
PROM and the NVRAM.
As stated earlier while the PROM acts as the interface for access to diagnostics and
drivers, the NVRAM consists of some editable user defined parameters. Non Volatile
information like the system identification information, device aliases etc are stored in
the NVRAM.The OpenBoot PROM is programmable and can be programmed based on Forth, which
is an interactive
programming language much like shell scripting.
The main tasks performed by the OpenBoot firmware are:
- Initializing and Testing system hardware ( POST , power on self test)
- Interactive Debugging
- Management of NVRAM Parameters
- Start the Operating System boot
Useful commands at OK prompt.
- this command shows the following systems
hardware informatiion : Model, architecture, processor,keyboard, openboot
version, Serial no. ethernet address & host id.
test floppy - test floppy disk drive
test net - test network loopbacks
test scsi - test scsi interface
test-all test for all devices with selftest method
- Show ticks of real-time clock
- Monitor network broadcast packets
- Monitor broadcast packets on all net
- Show attached SCSI devices
- Show attached SCSI devices for all host
adapters- internal & external.
boot - boot kernel from default device.
Factory default is to boot
from DISK if present, otherwise from NET.
boot net - boot kernel from network
boot cdrom - boot kernel from CD-ROM
boot disk1:h - boot from disk1 partition h
boot tape - boot default file from tape
boot disk myunix -as - boot myunix from disk with flags "-as"
- ok cd
- ok .properties
- ok ls
- f00809d8 tape
- f007ecdc disk
- ok .speed
- CPU Speed : 200.00MHz
- UPA Speed : 100.00MHz
- PCI Bus A : 66Mhz
- PCI Bus B : 33Mhz
Display all variables and current values.
- setenv <variable>
Set variable to the given value.
- set-default <variable>
- Reset the value of variable to the
- Reset variable values to the factory
- Key Sequences
- These commands are disabled if
the PROM security is on. Also, if your system has full security enabled,
you cannot apply any of the suggested commands unless you have the password to
get to the ok prompt.
- Stop - Bypass POST. This command does not depend on security-mode. (Note: some
systems bypass POST as a default; in such cases, use Stop-D to start
- Stop-A Abort.
- Stop-D - Enter diagnostic mode (set diag-switch? to
- Stop-F - Enter Forth on TTYA instead of probing. Use exit to continue with
the initialization sequence. Useful if hardware is broken.
- Stop-N Reset NVRAM contents to default values.
Start an OpenBoot Diagnostics
OK setenv diag-switch? true
setenv auto-boot? false
OK test-all or obdiag
Configure Graphics Console (e.g. Sun XVR-100 Graphics Accelerator) instead of serial
Select the graphics accelerator, e.g. b
OK nvalias mydev <CTRL-Y>
OK setenv use-nvramrc? true
If you are having problems where you have created a $HOME/.forward file in your home directory
to forward e-mails from one account to another and it just won't forward them?
First make sure the file isn't group or world writable.
-rwxrwxr-x 1 zahn dba 0 Jan 9 12:17 .forward # wrong
-rwxr-xr-x 1 zahn dba 0 Jan
9 12:17 .forward #
Lastly, make sure your home directory isn't group or world writable.
drwxrwxr-x 14 zahn dba 4096 Jan 9
12:20 zahn # wrong
drwxr-xr-x 14 zahn dba 4096 Jan 9 12:20 zahn # OK
A backup strategy is more complex than creating a redundant copy
of disk storage and considering the strategy a success. A successful backup strategy
must detail how the backup media are rotated, how the media are archived, how the
system will be recovered, and what the backup software will do to create the backup.
Although all parts of the backup strategy are equally important, this tip will focus on
the backup script and will detail a flexible backup script that uses built-in Solaris
software tools which create a reliable local backup of a Solaris machine.
The backup script will accomplish the following
- Create a backup archive that is as easy to restore a single
file as it is to restore an entire file system.
- The backup script will run autonomously. The only human
intervention will be to swap media and review output.
- The filesystems or directories to backup can be specified in
the script. Using automounter you can even specify remote filesystems.
- The script will create a detailed log of the
- The script will send an abbreviated email summary of the
backup to the administrator.
- After a successful backup, the script will verify to some
extent the contents of the backup media.
- The backup script will be able to run on any Solaris 2.6 or
greater machine without modification.
We use the well known utilities TAR, GZIP
and DD, because they are available on any Unix system. They are very well tested
and simple to use. In case of an emergency it is important to have a simple way to
restore, independent of complex tools and incompatible software releases.
The utility MT sends commands to a tape drive. Many of these commands are familiar,
but some are not. The script will use these mt commands.
- rewind – rewind the
- rewoffl – rewind the tape
and eject it (go offline)
- eom – space to end of
recorded media on tape
- weof – write count EOF marks
at current position on tape
- status – display current
status of tape
Click here for the Shell Script