Experimental Physics and
Industrial Control System

On 11/18/2014 10:48 AM, Andrew Johnson wrote:
> ...
> Not sure why this failed, it succeeds on my machine, but the output
> lines are in a different order and the "Message is" test line is missing:

It seems like the errlogFlush() on line 243 is returning before the
output of the preceding errlogPrintfNoConsole() has been seen by the
test's log listener.  Since pvt.count catches up it is clear that the
message isn't dropped, just late.

I can't fully explain the exact mechanism.  It's possible that the
short-circuit case of errlogFlush() is being hit.

> if (count <= 0)
>         return;

There might be a race if errlogFlush() runs after errlogThread() finds
msgbufGetSend()==NULL, but before it tests the flush event.

I'm not sure this could cause this specific behavior though.

References:

Jenkins build became unstable: epics-base-3.15 #143 APS Jenkins

Re: Jenkins build became unstable: epics-base-3.15 #143 Andrew Johnson

Navigate by Date:

Prev: Re: 3.15.0.2: Shutdown issues Benjamin Franksen

Next: Re: 3.15.0.2: Shutdown issues Michael Davidsaver

Index: 2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  2013  <2014>  2015  2016  2017  2018  2019  2020  2021  2022  2023 

Navigate by Thread:

Prev: Re: Jenkins build became unstable: epics-base-3.15 #143 Andrew Johnson

Next: Jenkins build is back to stable : epics-base-3.15 #144 APS Jenkins

Index: 2002  2003  2004  2005  2006  2007  2008  2009  2010  2011  2012  2013  <2014>  2015  2016  2017  2018  2019  2020  2021  2022  2023