On 2019/11/6 5:36 PM, Peter Levart wrote:
Hi Hamlin,
in TCPTransport.decrementExportCount():
283 try {
284 if (tcpLog.isLoggable(Log.BRIEF)) {
285 tcpLog.log(Log.BRIEF, "close server socket on
" + ss);
286 }
287 ss.close();
288 } catch (IOException e) {
289 }
...you could add a log statement to the catch block too. Or even
better, rearrange for IOException to be thrown from that method and
deal with it in two places:
- in exportObject() - add it as suppressed exception to exception
thrown from super.exportObject().
- in targetUnexported() - log it or wrap it into UncheckedIOException
(depending on what are the callers of targetUnexported())
What do you think?
Thanks Peter.
I agree. I adopt your first suggestion: add log statement to catch
block, as I think it's simple/straight and sufficient to help diagnose.
And I also add log for catch blocks in other close places.
The change is updated in place at:
http://cr.openjdk.java.net/~mli/8232446/webrev.00/
Thank you
-Hamlin
Regards, Peter
On 11/6/19 3:07 AM, Hamlin Li wrote:
Would you please review the patch?
bug: https://bugs.openjdk.java.net/browse/JDK-8232446
webrev: http://cr.openjdk.java.net/~mli/8232446/webrev.00/
We have some intermittent failures in rmi related to socket closing,
this is to add more logging to help diagnose the issues.
Thanks you
-Hamlin