Roll cycle possibly not picked up by tailer

[Jonathan H]

Hi,

Using Chronicle Queue 4.5.14, we have a long running tailer that outlives appender processes across roll cycles. So e.g. Monday morning the tailer starts and is still alive on Tuesday, while the appender processes start again on Tuesday morning after being shutdown Monday evening before midnight. Once the appender starts writing data to a new roll cycle (Tuesday) it seems like the tailer does not always catch that new data being written.

I have attempted to create a test replicating this behaviour: https://gist.github.com/jheusser/28621f722f0afa6137e80ef0e0afc276

If you uncomment line 115 where the appender writes data on day 1 the tailer picks up all data and the test passes, without it it's not reading anything and the test fails. 

Thanks

Jonathan

[Jonathan H]

Actually having to add, the test fails because theres 0 documents read, not 1 just one and the assertion fails normally.

[Rob Austin]

Could you raise this in chronicle queue GitHub issues and we will take a look. 

Thanks

Rob

Sent from my iPhone

--
You received this message because you are subscribed to the Google Groups "Chronicle" group.
To unsubscribe from this group and stop receiving emails from it, send an email to java-chronicl...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Jonathan H]

Thanks Rob, I raised an issue on github.

[Rob Austin]

thanks