Solaris

Enabling multipathing on Solaris 10

Solaris 10 does a pretty good job of handling multipath functionality with very little help from the systems administrator. Having Solaris use multipathing is as simple as running the stmsboot -e command. This will ask you a few questions about which controllers to enable it on if you are sure and if you want to go ahead and reboot to make the changes take effect. This will reboot the box, detect all multipathing settings, set them, and reboot once again in a new glorious multipath enabled world.

Synchronizing filesystems with rsync

Synchronizing multiple boxes can be important for multiple reasons, high-availability, disaster recovery, poor mans backups, etc... Here is a quick tutorial on using rsync to synchronize a couple of boxes. In this example one box is the master one is the slave. WARNING, this approach isn't the most secure... you should look into the auth users option, especially if you are doing this on the Internet and not your internal network.

On the slave, create a /etc/rsyncd.conf file with these contents:
[mirror1]
path = /place/to/mirror/to1
uid = SYNCUSER
gid = SYNCGROUP

Setup NTP on Solaris

Configuring NTP on Solaris is pretty trivial, first create the ntp.conf file in /etc/inet/ntp.conf it should look like the following with your information instead:
#ntp.conf
server ntp1.yourdomain.com
server ntp2.yourdomain.com

driftfile /etc/ntp.drift

Then you need to enable the service:
svcadm enable ntp

Updating a solaris box without X Windows

Sometimes either out of necessity or preference you may find yourself trying to patch a Solaris box without using x windows. Here is a quick guide to get you through:

First you have to register the system with Sun by providing your sun login credentials to get access to patches that can be done with the following
cp /usr/lib/breg/data/RegistrationProfile.properties /tmp/

Fill in your account information in the properties file (replace the caps with your data):
perl -pi -e 's/^userName=/userName=YOURNAME/' /tmp/RegistrationProfile.properties

Chown or chgrp fails across nfs mount points

If you experience chown failures where you have no_root_squash (root= in /etc/dfs/dfstab) specified for a nfs client you are probably dealing with an issue of named user/group passing in NFSv4 vs id based passing in NFSv3. Your options are to either specify version 3 or make sure that the users/groups exist on both the cilent and server or make sure they don't exist on the client. That will ensure the id is passed as numeric as there is no name to lookup and pass.

Example:
/etc/group excerpt on client (group monkeys doesn't exist on NFS server)
monkeys::101:

NFS exports on solaris

I have been using Linux so long that I constantly forget how to do the same operations on a Solaris box, and NFS is definitely different. So here's a note so I don't forget, hope it's useful to someone else.

Linux (redhat/centos):
edit /etc/exports
/etc/init.d/nfs reload

Solaris:
edit /etc/dfs/dfstab
shareall

Syndicate content