Display or change default polling parameters of a node (QNX)
netpoll [-i idle_time] [-n node] [-p poll_period]
[-r num_retries]
- -i idle_time
- The length of time, in seconds, that the system will allow a
virtual circuit (VC) to remain idle
before checking the VC's integrity (default is to not change
the idle time from its current setting). The integrity check
consists of transmitting a network poll packet to the remote node.
- -n node
- The node on which to change the poll parameters. The default is the node
from which the utility is invoked.
- -p poll_period
- The length of time, in seconds, between checks for idle VCs
once the initial idle_time (see -i) has
elapsed (default is to not change the poll period).
- -r num_retries
- The number of times that the system will transmit a network poll packet
to a remote node before it concludes that the remote node is permanently
unreachable (default is to not change the number of retries).
In QNX, virtual circuits (VCs) provide transparent
interprocess communication between processes residing on
two different nodes in a network. If a node abruptly powers down,
or if it can no longer be communicated with, the resources associated
with the VCs to that node should be recovered.
Each node polls its own VCs, checking their integrity. The
netpoll utility lets you view or change the default polling
parameters of a specific node.
To display the polling parameters for a node, don't specify any parameters;
if the node is remote, specify only the
-n node option.
|
The initial polling parameters are determined by the version of
Net that is being run. Typical initial values are:
Parameter |
Value |
poll period |
10 seconds |
idle time |
10 seconds |
num retries |
6 |
|
The following example:
- reduces to an average of 10 seconds the time allowed to detect a node
"disappearing"
(poll_period * num_retries),
at the expense of increased CPU consumption
- increases the frequency of the VC polling
- decreases the maximum number of allowed polls:
netpoll -p 2 -i 2 -r 5
The following example:
- increases to an average of 200 seconds the time allowed to detect
a node "disappearing"
(poll_period * num_retries)
- decreases the frequency of the VC polling
- increases the maximum number of allowed polls:
netpoll -p 20 -i 20 -r 10
Net,
netboot,
netinfo