Personal tools
You are here: Home Support and Documentation Administrator's Guide 6.6.  ThinLinc HA and Other Services
Document Actions

6.6.  ThinLinc HA and Other Services

6.6.  ThinLinc HA and Other Services

Since no chain is stronger than its weakest link, it's important to think about all services that are needed for proper ThinLinc operations when planning a HA system. For example, is the user database, for example an eDirectory server, fault-tolerant? If not, a HA setup of ThinLinc will not help when the eDirectory fails.

In this section, we will give some hints on things that need to be considered for proper operations in case of failure.

6.6.1.  User database

The database of users that should be allowed to login to the ThinLinc cluster must not be sensitive to hardware failures, or users will not be able to login to ThinLinc. If the database of users are in the form of a winbind/NIS combination as described in Section 9.3, “ Authenticating users against Windows Active Directory ”, samba and the YP server must be started by heartbeat. Help with doing this is available on a consulting basis from Cendio AB.

6.6.2.  Home Directories and other File Resources

The home directories and other file resources used by the ThinLinc cluster must also be fault-tolerant, or ThinLinc will not be usable for the users. This is also outside the scope of ThinLinc HA.

6.6.3.  Printing

Depending on the needs of your organisation, it might be acceptible that printing is down for a while, if a printing server has a hardware failure.

Regarding the ThinLinc printer features nearest and thinlocal, thinlocal is normally not affected since it's defined on each VSM agent server. It does need to contact a VSM server to fetch some information, but since the variable /vsmagent/master_hostname is set to the HA hostname, it will always find a VSM server as long as any of the two VSM server nodes are up and running.

nearest is normally defined on one CUPS server in a ThinLinc Cluster. When running in HA mode, nearest must be be defined on both VSM server nodes, which should ensure operations even if the primary has failed. The database with mappings between client hardware (MAC) adresses and printer locations must be replicated between both nodes. If there are problems with duplicated nearest printers on the VSM agents, try running CUPS as a HA service via Heartbeat.