Aleksey Plekhanov created IGNITE-18307:
------------------------------------------
Summary: Long wait on locHost.isReachable
Key: IGNITE-18307
URL: https://issues.apache.org/jira/browse/IGNITE-18307
Project: Ignite
Issue Type: Improvement
Reporter: Aleksey Plekhanov
Assignee: Aleksey Plekhanov
When node starts it submit background task to check if local host is reachable
(see {{{}GridDiagnostic#runBackgroundCheck{}}}). This check only prints warning
to the log in some cases, but execution of {{locHost.isReachable}} method can
be relatively durable (for example, on my laptop it's executed for about 1
second, on TeamCity server looks like it's executed for about 500 ms). This
task executed asynchronously, but before node stops it shut down pools and
waits for this task, so start and stop of the node can't consume less time
than \{{locHost.isReachable}} method.
We have some tests, which are start the node, do some trivial work and stop the
node in cycle or with different parameters. Currently each cycle of such a
tests can consume more than one second, but actially start/stop node without
\{{locHost.isReachable}} and performing trivial action consumes tens or
hundreds of milliseconds.
Proposal: do \{{locHost.isReachable}} check only once per JVM instance, this
can reduce a little time of tests run on TeamCity.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
