hadoop-zookeeper-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Benjamin Reed <br...@yahoo-inc.com>
Subject RE: multiple disconnect events is not a state change.
Date Mon, 05 Jan 2009 22:28:19 GMT
it is a bug.

ben
________________________________________
From: Patrick Hunt [phunt@apache.org]
Sent: Monday, January 05, 2009 2:17 PM
To: Patrick Hunt
Cc: zookeeper-user@hadoop.apache.org
Subject: Re: multiple disconnect events is not a state change.

Sorry, meant "current client lib code"

Patrick Hunt wrote:
> I see what you are saying now, good question. :-)
>
> In looking at the current server code we generate a disconnected state
> event every time the client lib attempts to connect to the server and
> fails, even if we are currently in the "disconnected" state.
> Ben/Flavio/Mahadev, any insight into why this is? Is this done for a
> reason or a "bug"?
>
> Patrick
>
> Kevin Burton wrote:
>> oh... I was saying that I was receiving an event.  I was just printing
>> it to
>> stdout....
>> Should there be multiple disconnect events?  There probably shouldn't
>> be...
>> you can only be disconnected once.
>>
>> Kevin
>>
>> On Mon, Jan 5, 2009 at 1:29 PM, Patrick Hunt <phunt@apache.org> wrote:
>>
>>> What you are seeing is the client attempting to reconnect to a server in
>>> the server list after a period of time (or retry if there is only a
>>> single
>>> server host in the list). There has been a patch since the last release
>>> which logs an INFO level message during client connection attempt, so
>>> this
>>> should be more obvious in future (you will see the disconnected but
>>> you will
>>> also see a new connection attempt message as well).
>>> https://issues.apache.org/jira/browse/ZOOKEEPER-225
>>>
>>> However this is not output to stdout, please feel free to enter a
>>> jira for
>>> this.
>>>
>>> Regards,
>>>
>>> Patrick
>>>
>>>
>>> Kevin Burton wrote:
>>>
>>>> Shutting down my zookeeper server yields this on my client.
>>>>
>>>> Continual disconnect events.
>>>>
>>>> Shouldn't only one be issued?  The second one is not a state change.
>>>>
>>>> WatchedEvent: Server state change. New state: Disconnected
>>>> WatchedEvent: Server state change. New state: Disconnected
>>>> WatchedEvent: Server state change. New state: Disconnected
>>>> WatchedEvent: Server state change. New state: Disconnected
>>>> WatchedEvent: Server state change. New state: Disconnected
>>>> WatchedEvent: Server state change. New state: Disconnected
>>>> WatchedEvent: Server state change. New state: Disconnected
>>>>
>>>>
>>>>
>>
>>

Mime
View raw message