RFR 8085875/10, java/net/DatagramSocket/PortUnreachable.java fails intermittently: Address already in use

[Felix Yang]

Hi Mark and Christoph,

      thanks for the comment.

     Pushed with changing max tries count into a "staic final int" field 
and removed the in-necessary space character.

-Felix
On 2017/9/7 22:04, Langer, Christoph wrote:
> Hi Mark,
>
> I overlooked that, you are completely right, recreateServerSocket is fine then.
>
> Best regards
> Christoph
>
>> -----Original Message-----
>> From: Mark Sheppard [mailto:mark.sheppard at oracle.com]
>> Sent: Donnerstag, 7. September 2017 13:07
>> To: Langer, Christoph <christoph.langer at sap.com>; Felix Yang
>> <felix.yang at oracle.com>
>> Cc: net-dev at openjdk.java.net
>> Subject: Re: RFR 8085875/10,
>> java/net/DatagramSocket/PortUnreachable.java fails intermittently: Address
>> already in use
>>
>> Hi Christoph,
>>      looking at the logic of the test, it is seen that a "server socket"
>> is created in the execute method, and is then closed,
>> and a bunch of datagrams sent to it.  Then in the serverSend method an
>> attempt is made to (re-)create a DatagramSocket using the
>> same port as the original "server socket", so recreate would seem to
>> convey an appropriate semantics.
>> WRT an upper limit for retryCount, that's a reasonable point, but
>> probably a "static final int" rather than just static field.
>>
>> regards
>> Mark
>>
>> On 07/09/2017 10:40, Langer, Christoph wrote:
>>> Hi Felix,
>>>
>>> this looks good in general.
>>>
>>> I would, however, suggest to rename the method 'recreateServerSocket'
>> into ' createServerSocket' as the former name suggests that something
>> which existed once was recreated. But in this case the socket is simply
>> created with a few retries when exceptions occur.
>>> I'd also prefer if the retry count could be stored in a static field to allow for
>> easy modification, instead of literally coding it to the value of 5 in the
>> method.
>>> And a minor thing: in line 63 there's a blank in between
>> "recreateServerSocket" and "(int serverPort)" which should be removed.
>>> Best regards
>>> Christoph
>>>
>>>> -----Original Message-----
>>>> From: net-dev [mailto:net-dev-bounces at openjdk.java.net] On Behalf Of
>>>> Felix Yang
>>>> Sent: Mittwoch, 6. September 2017 10:06
>>>> To: net-dev at openjdk.java.net; Mark Sheppard
>>>> <mark.sheppard at oracle.com>
>>>> Subject: RFR 8085875/10,
>> java/net/DatagramSocket/PortUnreachable.java
>>>> fails intermittently: Address already in use
>>>>
>>>> Hi all,
>>>>
>>>>        please review a patch to add retries to avoid possible port
>>>> conflicts during re-binding.
>>>>
>>>> Bug:
>>>>
>>>>        https://bugs.openjdk.java.net/browse/JDK-8085875
>>>>
>>>> Webrev:
>>>>
>>>>        http://cr.openjdk.java.net/~xiaofeya/8085875/webrev.00/
>>>>
>>>> Thanks,
>>>>
>>>> Felix