Create a Soalris 8 zone (32bit) to a container

Last time I had successfully created a flar image from an old Solaris 8 server (ref). Now I have to setup a zone of Solaris 10 container and install the image into it. The overall idea is clear and easy. Normally just create the zone configuration, and install it from a "flar" image.

1. Create zone configuration

root@sol10 # zonecfg -z z_sol8
z_sol8: No such zone configured
Use 'create' to begin configuring a new zone.
zonecfg:z_sol8> create -t SUNWsolaris8
zonecfg:z_sol8> set zonepath=/pool/zone/sol8
zonecfg:z_sol8> set autoboot=false
zonecfg:z_sol8> set ip-type=shared
zonecfg:z_sol8> add net
zonecfg:z_sol8:net> set address=192.168.123.123
zonecfg:z_sol8:net> set physical=nxge0
zonecfg:z_sol8:net> end
zonecfg:z_sol8> verify
zonecfg:z_sol8> commit
zonecfg:z_sol8> exit

Next install the zone from the flar image:       

Failed to exclude directories when running flarcreate on Solaris 8

When I tried to prepare a flash archive (flar) from an old Solaris 8 physical machine, I wanted to exclude two directories (say, /one and /two). I checked the syntax from web (and the man page), it should be:

flarcreate -n -c -S -R / -x /one -x /two
where -c is enable compression, -S is ignore size stats, -R is the root path

However, the directories /one and /two are still in the flar !

After doing some investigation and research from the web, it is believed that there is a bug of this script of the current version. Since it's too old and I don't want to apply any patch on it, I'd rather modify the script a bit in order fulfill my requirement:

Check utilization for Solaris 10

Seems "top" is not available (or not installed by default). Actually there is another command "prstat" which provide similar but more powerful functions:

#prstat -Z
   PID USERNAME  SIZE   RSS STATE  PRI NICE      TIME  CPU PROCESS/NLWP
  5147 root      206M   33M sleep   59    0   0:03:10 0.0% webservd/30
  1378 noaccess  157M  129M sleep   59    0   0:02:41 0.0% java/18
  2333 root      130M   90M sleep   59    0   0:07:52 0.0% java/72
  5060 root       82M   14M sleep   59    0   0:03:11 0.0% webservd/71
  5088 106        81M   20M sleep   59    0   0:03:11 0.0% webservd/72
  1865 root       63M   24M sleep  101    -   0:00:04 0.0% rgmd/43
  1009 root       51M   24M sleep   59    0   0:00:22 0.0% fmd/30
  1719 root       48M 4080K sleep  100    -   0:00:19 0.0% rpc.pmfd/23
  1119 root       44M   28M sleep   59    0   0:34:05 0.0% poold/8
    11 root       36M   29M sleep   59    0   0:00:30 0.0% svc.startd/13
  5094 root       34M   16M sleep   59    0   0:00:50 0.0% opceca/2
   861 root       28M 3520K sleep   59    0   0:00:00 0.0% pmmd/20
  5146 root       26M 3344K sleep   59    0   0:00:07 0.0% webservd/2
  5087 root       26M 2976K sleep   59    0   0:00:07 0.0% webservd/2
  5059 root       26M 2600K sleep   59    0   0:00:07 0.0% webservd/2
ZONEID    NPROC  SWAP   RSS MEMORY      TIME  CPU ZONE
     0       95  734M  566M   0.9%   1:05:10 0.0% global
     1       37  275M  112M   0.2%   0:10:00 0.0% zone1
     3       44  113M  112M   0.2%   3:18:50 0.1% zone2
     2       23   18M   26M   0.0%   0:01:21 0.0% zone3

Total: 199 processes, 4159 lwps, load averages: 0.43, 0.39, 0.38

Note that I run this command one a container with 3 zones. The -Z parameter provide a glance of the utilization of each zone. To view the utilization of an individual zone, you can run "prstat -z ":

Change network speed and duplex mode on Solaris 10

When I try to FTP some files using an interface, the speed is unexpectedly slow. Then I try to see the network interfaces status:

root@sunserver # dladm show-dev
nxge0           link: down      speed: 0     Mbps       duplex: unknown
nxge1           link: down      speed: 0     Mbps       duplex: unknown
nxge2           link: unknown   speed: 0     Mbps       duplex: unknown
nxge3           link: unknown   speed: 0     Mbps       duplex: unknown
igb0            link: up        speed: 100   Mbps       duplex: half
igb1            link: unknown   speed: 0     Mbps       duplex: half
igb2            link: up        speed: 100   Mbps       duplex: full
igb3            link: up        speed: 1000  Mbps       duplex: full

Seems 100 Half is a wrong setting. Since the speed is set by auto negotiation mode, we have to change it manually:

First, check the interface properties by "ndd":
root@sunserver # ndd -get /dev/igb0 ?
?                             (read only)
mtu                           (read and write)
min_allowed_mtu               (read only)
max_allowed_mtu               (read only)
adv_autoneg_cap               (read and write)
adv_1000fdx_cap               (read and write)
adv_1000hdx_cap               (read only)
adv_100fdx_cap                (read and write)
adv_100hdx_cap                (read and write)
adv_10fdx_cap                 (read and write)
adv_10hdx_cap                 (read and write)
adv_100T4_cap                 (read only)
link_status                   (read only)
link_speed                    (read only)
link_duplex                   (read only)
autoneg_cap                   (read only)
pause_cap                     (read only)
asym_pause_cap                (read only)
1000fdx_cap                   (read only)
1000hdx_cap                   (read only)
100fdx_cap                    (read only)
100hdx_cap                    (read only)
10fdx_cap                     (read only)
10hdx_cap                     (read only)
lp_autoneg_cap                (read only)
lp_pause_cap                  (read only)
lp_asym_pause_cap             (read only)
lp_1000hdx_cap                (read only)
lp_1000fdx_cap                (read only)
lp_100fdx_cap                 (read only)
lp_100hdx_cap                 (read only)
lp_10fdx_cap                  (read only)
lp_10hdx_cap                  (read only)
link_autoneg                  (read only)
tx_copy_thresh                (read and write)
tx_recycle_thresh             (read and write)
tx_overload_thresh            (read and write)
tx_resched_thresh             (read and write)
rx_copy_thresh                (read and write)
rx_limit_per_intr             (read and write)
intr_throttling               (read and write)
adv_pause_cap                 (read only)
adv_asym_pause_cap            (read only)

As those "read only" properties are not useful, let's filter out further:

SUN Cluster (3.3) installation screen dump

Just a screen dump of the installation. Assumed the Solaris 10 is installed and media of the cluster is ready.

---

Just a screen dump of the installation. Assumed the Solaris 10 is installed and media of the cluster is ready.

# cd 
# ./installer

Unable to access a usable display on the remote system. Continue in command-line mode?(Y/N)

Y

Java Accessibility Bridge for GNOME loaded.

 

 

Clone a new machine by spliting a miror disk (Vertias VM encapsulated)

Situation: suppose you are going to clone the machine A to machine B

• Before cloning, make sure the disks are in sync (vxprint -htA, format, etc..)
• Machine A and Machien B: Shutdown to OK prompt (init 0)
• Removal of disks should be in OK prompt
• Insertion of disks can be in OS level
  

Machine A:

1. Remove one of the disk and try to boot it with single in OK prompt (boot )

2. after the OS is up, vxprint should shown the another disk is missing/fault
3. insert a new disk and issue "vxdctl enable"

4. vxdiskadm -> 5

5. resync should be in progress, use "vxtask list" to monitor

6. Double check (vxprint -htA) and format, if the partition label is not correct, issue "/etc/vx/bin/vxbootsetup (the cxtxdx doesn't work. use 'vxdisk list' to get the disk name first)"

7. Try to reboot once again

Machine B:

1. insert the disk of Machine A and boot in OK prompt (boot )

2. after the OS is up, vxprint should shown the another disk is missing/fault
3. insert a new disk and issue "vxdctl enable"

4. vxdiskadm -> 5

5. resync should be in progress, use "vxtask list" to monitor

6. Double check (vxprint -htA) and format, if the partition label is not correct, issue "/etc/vx/bin/vxbootsetup (the cxtxdx doesn't work. use 'vxdisk list' to get the disk name first)"

7. Try to reboot once again

How to add a swap file in Solaris

1. # mkfile 10m /tmp2/swapfile
2. # swap -a /tmp2/swapfile
3. # swap -l

   • swapfile             dev  swaplo blocks   free
     /dev/dsk/c0d0s1     102,1       8 303400 303400
     /dev/dsk/c0d1s6     102,70      8  21160  21160
     /tmp2/swapfile        -        8  20472  20472
       

4. (Optional, if need to make the change permanent)
   1. Edit /etc/vfstab

              /tmp2/swapfile  -       -       swap    -       no      -
        

   2. reboot