qb_domain
Synopsis
qb_domain = qube_domain_name
The qb_domain value is distinct and separate from the DNS domain name (e.g. pipelinefx.com) or an ActiveDirectory domain name.
This is a Worker startup parameter. Since this parameter is involved in the initial startup of the worker and affects the behavior of the worker before it contacts the supervisor to request a "remote" configuration from the central worker configuration file (see qbwrk.conf), it's considered a bootstrapping parameter and can only be specified in the worker's local qb.conf configuration file. Any mention in the global qbwrk.conf file will be ignored.
This is a complete list of the Worker bootstrap parameters
- qb_directory
- qb_domain
- qb_supervisor
- worker_address
- worker_boot_delay
- worker_journal_location
- worker_lookup
- worker_max_threads
- worker_pidfile
- worker_port
Usage Notes
The qb_domain is best left to the default value unless the administrator intends to set up 2 or more farms, in which case only the smaller of the two farms should use a non-default value.
A Qube! Supervisor will not respond to any request or transaction from a worker or client whose qb_domain value does not match its own.
If a Supervisor receives a worker registration request (such as when a new worker first starts up on the farm), but detects a mismatch between the workers and Supervisor's qb_domain value, the worker's registration request is rejected, and the worker will attempt to find another supervisor to bind to. If there are no more supervisors to reply to the worker, the worker will never appear in any supervisor's worker list.
If the request is a from a Qube! client, the request is ignored, and the client will attempt to locate another supervisor. If no supervisors is found whose qb_domain value matches the client's, the client will usually issue an:
ERROR: udp receive socket timed out.
error message.
About supervisor auto-discovery
Since auto-discovery uses network broadcast packets, and these packets are not forwarded by routers, supervisors can only be discovered if they are on the same IP subnet as the host issuing the broadcast.
Qube workers that don't have the supervisor explicitly defined via the qb_supervisor value can issue a broadcast across the network and ask any supervisor that's running to respond. This is normally sufficient unless there are multiple supervisors on the network that can respond to the broadcast (on the same IP subnet as the worker). In this case, the qb_domain value provides a check for the auto-discovery mechanism: the qb_domain value is compared between the supervisor and the worker, and the supervisor only responds to requests from workers whose qb_domain value matches the supervisor's.
Example
qb_domain = MyDomain
Defaults
qb_domain = qube
It's inadvisable to vary from the default qb_domain value for the main part of your farm.
The reason to stick with the defaults is that if you deviate from the default qb_domain value on the Supervisor, you have to set it on every client and worker in order to achieve proper operation. Any new machines you add will require this configuration change before they will work properly. Deviating from the default doesn't provide any advantages, and it's one more thing to miss during new host setup.