Re: [xmlblaster] Fail safe mode/invalid secretsessionId

[Michele Laghi <michele at laghi.eu>]

Hi Xavier,

Xavier Roques wrote:
> Hi Michele,
> 
> Thanks for your quick answer.
> 
> I think that server closes the session because of a ping fail (dispatch/callback/retries=0 is set).
> 
ok, but you should have:
dispatch/callback/retries=-1

What you describe makes sense. We have worked heavily in the current svn
to get rid of the problems we had with the pings (callback and
connection). Particularly pings could fail if:
- an ongoing operation was blocking the socket (for example slow
connections or big data)
- the processing of the command was particularly slow (CPU 100% on the
server).

If you had one of these scenarios I suggest you take the svn code.

> There is no timeout on the session and there are no JMX calls.
> 
> In the log of the agent we can see the following lines:
> 2006-08-04 11:35:39,725 WARN  [Thread-4] xmlBlaster.dispatch - [DispatchConnection-connection:client/mwec9a0205-agent/10] initialize:XmlBlasterException errorCode=[communication.noConnection] serverSideException=false location=[SocketUrl] message=[#1.1 SSL XmlBlaster server is unknown, '-dispatch/connection/plugin/socket/hostname=<ip>' : java.net.ConnectException: Connection refused: connect] [See URL http://www.xmlblaster.org/xmlBlaster/doc/requirements/admin.errorcodes.listing.html#communication.noConnection]
> 
> 2006-08-04 11:35:39,766 WARN  [Thread-4] xmlBlaster.dispatch - [DispatchConnection-connection:client/mwec9a0205-agent/10] Connection transition UNDEF -> POLLING: socket://msilsa0215:7609 is unaccessible, we poll for it every 2000 msec: errorCode=communication.noConnection message=#1.1 SSL XmlBlaster server is unknown, '-dispatch/connection/plugin/socket/hostname=<ip>' : java.net.ConnectException: Connection refused: connect
> 
> 2006-08-07 15:40:12,371 ERROR [Login requestor mwec9a0205-agent] errorCode=user.security.authentication.accessDenied message=#1.1 Your secretSessionId is invalid, no access to xmlBlaster_162_111_89_158_3412.
> 
> After the error " user.security.authentication.accessDenied " there are no more lines into the agents. If it has been disconnected, I should see lines
> Like this one no?
> 2006-07-26 17:30:55,205 WARN  [XmlBlaster.PingTimer] xmlBlaster.dispatch  [DispatchConnection-connection:client/mwec9a0205-agent/10] No connection established, socket://msilsa0215:7609 still seems to be down after XXX connection retries
> 
> Unfortunately, it is not easily reproducible and most of the time all is ok, but in some rare cases some agents are not able to connect to the server.
> 
> Currently, I updated the code of our agents in order to reconnect if such an exception occurred. But I don't understand why some agents do not know that they lose the connection with the server (They are using a killed session). 
> 
You can create the error by killing the session with jmx
Regards
Michele

> Best regards,
> 
> Xavier.
> 
> 
> 
> -----Original Message-----
> From: owner-xmlblaster at server.xmlBlaster.org [mailto:owner-xmlblaster at server.xmlBlaster.org] On Behalf Of Michele Laghi
> Sent: jeudi 17 août 2006 14:10
> To: xmlblaster at server.xmlBlaster.org
> Subject: Re: [xmlblaster] Fail safe mode/invalid secretsessionId
> 
> Hi Xavier,
> what you described happens when the server has no session anymore (for
> example if it has been killed for some reason) and the client does not
> know about it and tries to work as usual (publish subscribe etc.)
> 
> Xavier Roques wrote:
>> Hello,
>>
>> After doing an OS patch on some of our boxes
>> We notice that some agents were not connected to the xmlblaster server.
>>
>> In the log of the agents we can see the following errors:
>>
>> INFO   | jvm 1    | 2006/08/07 15:40:12 | 2006-08-07 15:40:12,371 ERROR
>> [Login requestor mwec9a0205-agent] event.LoginRequestor
>> (com.infovista.foundation.jmx.event.LoginRequestor:LoginRequestor.java:r
>> un:87)     - com.infovista.foundation.middleware.mom.MOMException:
>> errorCode=user.security.authentication.accessDenied message=#1.1 Your
>> secretSessionId is invalid, no access to xmlBlaster_162_111_89_158_3412.
>>
>> And the agent does not try anymore to connect to the Xmlblaster server.
>>
>> We are using xmlblaster 1.1
>> All our agents are configured to use the fail safe mode and only one
>> session of each agent is allowed ("sameclientonly").
>> The protocol used is the SOCKET one.
>>
>> How can I fix this issue?
>>
> 
> I think it is the job of the client application (in your case your
> agent) to decide what to do if such an exception happens. In your case
> you would wish to reconnect (explicit reconnect).
> 
> The underlying problem, that is the reason why the server has killed the
> session is unknown to me: I would need some more details, for example
> configuration and server logs.
> 
> Some of the reasons could be because:
> - explicit killing of the session (for example via jmx)
> - session timeout
> - callback ping failing and no -dispatch/callback/retries=-1 set
> - ...
> 
> Regards
> Michele
> 
>> Thanks,
>>
>> XavieR.
>>
>>
>>
> 
> 
> 
>