river-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Patricia Shanahan <p...@acm.org>
Subject RetryTask test coverage
Date Wed, 28 Jul 2010 17:57:24 GMT
com.sun.jini.thread.RetryTask is a commonly used implementation of 

It retries a subclass-supplied method at a series of intervals, ranging 
up to 5 minutes. That seems a good basic plan. The different intervals 
would allow for different forms of failure and recovery. 5 minutes is 
long enough that an application that implements a service could have 
failed, exited, been restarted by a keep-alive script, and got back to 
work within the interval.

However, I am concerned that there are situations involving runAfter 
that are much more probable with a long delay than if the RetryTask 
succeeds the first time. In local test environments, there may be a 
tendency to succeed on the first try more often than would be the case 
in the wild.

My question is whether the tests cover the longer delays. It is 
answerable, if nobody already knows the answer, from a "FINEST" level 
log for com.sun.jini.thread.RetryTask. If RetryTask is doing retries, it 
will contain messages of the form "retry of {0} in {1} ms". {0} is the 
toString for an instance of a RetryTask subclass. {1} is the time in 
milliseconds to the next retry. If possible, I would like to see those 
messages for a full test sequence.



View raw message