JDBC Receiver Channels stuck !!!!

Hello,

We are facing a very strange situation..

From our SAP/R3 system we are sending data to 3 databases (3 dif. Countries). And the three of them have different verison of Oracle.

9.2.0.6 , 9.2.0.8 and 10.2.0.4.

The problem happens only with channels that are sending to 9.2.0.8 and not all the time.. The problem started after one month from Go Live and now it happens almost one time every week with exception previous week that happend twice the same day!!

In all channels we have set the following advanced parameters

taskTimeout :120

driver:oracle.net.CONNECT_TIMEOUT : 120000

driver:oracle.jdbc.ReadTimeout : 120000

SAP Managed Tags:
SAP Process Integration

You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.

Comment

Accepted Solutions (0)

Answers (2)

Hi John Banellas,

We are facing the same situation. Did you find any better solution?

Thank's

You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.

Comment

John,

What do you mean by "stuck"? Please elaborate.

Regards,

Neetesh

You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.

Comment

Sorry, I pressed post by mistake before finishing..

When in XI the channel is stuck, in message monitoring we have 3 messages with status delivering and all others are with status to be delivered and it will not start over again unless I restart the node that Java is.

There is no error and in Communication Channel Monitoring the channel is Green and Functioning.

Is there any other way to restart the channel without restarting XI?

Of course the best solution is to fix this issue.

Thank you in advance,

John B.


Is there any other way to restart the channel without restarting XI?

There is no solution for getting rid of "delivering" messages other than re-starting the java node or the server itself. But why you re-starting the XI? You can simply re-start the java node on which the message is stuck. It's pretty quick and doesn't hurt any running process ...

Hope this helps.

Regards,

Neetesh

Well I think that our Basis are doing what you said.

They are logging to the server that Java is on and they stop/start the dispatcher and takes about 10 minutes..

But that is not an option because when they do that no message is coming or going also to the other databases (Countries), and ecxept those connections we have HTTP connection with our stores.. So.. even that 10 minutes could cause an issue.

Also whenever that occurs I have to find/call a Basis in order to do it..

Thanks,

John B.

My apologies for confusing you here - what I meant in my earlier post was the cluster node and not Java stack. You can re-start the cluster node where the message is stuck, it won't affect other messages and the re-start takes 5 minutes.

But if you don't have cluster nodes, then re-start of Java stack is the only option left.

Basically you go through these two threads to find a permanent solution for this issue..

I am trying to post a reply but displays like s...... :S

Edited by: John Banellas on Aug 9, 2010 3:30 PM

Edited by: John Banellas on Aug 9, 2010 3:36 PM


> My apologies for confusing you here - what I meant in my earlier post was the cluster node and not Java stack. You can re-start the cluster node where the message is stuck, it won't affect other messages and the re-start takes 5 minutes.
> 
> But if you don't have cluster nodes, then re-start of Java stack is the only option left.

Hi Neetesh,

We have cluster nodes. And when this happens (like 10 minutes ago ) our basis are going to the node that are the messages and they are stop/start only the dispatcher. When they do that we have problems with java but they are on it.


> Basically you go through these two threads to find a permanent solution for this issue..
> 
> 
> 
>

Hi Sarvesh,

I went through those threads.. most of what they are saying we have implemented them...

.maxReceivers has been set up to 3 with deafaultMaxConsumersPerQueue 10...

MaxThreadCount has been set up to 500 and in Thread manager InitialThreadCount has been setup to 80, MinThreadCount to 80 and MaxThreadCount to 120..

Before setting the above all the Receiver Channels was stuck!!

What we haven't set is

1) poolWaitingTime which I did it know and I will wait to see what will happen.

2) sqlquerytimeout which to be honest I didn't quite understood what is for.. and

3) messaging.connections which are

Send.maxConsumers=5

Recv.maxConsumers=5

Call.maxConsumers=10

Rqst.maxConsumers=10

In one post from the threads you give says that they should be 10,10,20,20

For now I am waiting to see the next dayu2019s how the system will behave with the poolWaitingTime that I have setup.

thanks for your answers, and I will let you know.

BR,

John B.


> 1) poolWaitingTime which I did it know and I will wait to see what will happen.

Hello,

Unfortunately the poolWaitingTime didn't work.. I tried 30000 mlseconds and 120000 mlseconds..

From Monday till Today once per day the channel was stuck and I was informed that in the country that is the specific database they had network issues!!! Today for 20 minutes ( I should put poolWaitingTime for more than 20 minutes?)

For sure it is not normal when they JDBC cannot connect not to be able to connect when the network is ok..

What I should try before creating a message for SAP ??

Thanks,

John B.

John,


Unfortunately the poolWaitingTime didn't work.. I tried 30000 mlseconds and 120000 mlseconds..
 
From Monday till Today once per day the channel was stuck and I was informed that in the country that is the specific database they had network issues!!! Today for 20 minutes ( I should put poolWaitingTime for more than 20 minutes?)

Pool Wait time of 20 minutes is a lot, and I won't recommend you for that.

Before opening a message with SAP, please make sure the network connectivity with the DB is perfect because if that's the case, even SAP won't be able to help

Regards,

Neetesh


> > Before opening a message with SAP, please make sure the network connectivity with the DB is perfect because if that's the case, even SAP won't be able to help 
> 
> Regards,
> Neetesh

Neetesh,

network connectivity, especially when it is between countries, is something that you could have problems any time..

One of the reasons to use XI was that even if the other end is down, XI will push or pool data from it when will go up..

Restarting the node, or Java or whatever, everytime the other end is down is not an option..

So, from my point of view SAP should be able to help especially in that case...

BR,

John B.

John,


network connectivity, especially when it is between countries, is something that you could have problems any time..
One of the reasons to use XI was that even if the other end is down, XI will push or pool data from it when will go up..

I don't completely agree with you - if the DB is down, then message will be in PI in error state and once the DB is back, you can re-send the message.

But if the message gets into "delivering" state, it is not because of receiving system being "down", but due to a network glitch.


Restarting the node, or Java or whatever, everytime the other end is down is not an option..

I completely agree with you on this. Whenever we face this issue, we had to re-start the cluster node. We had tried all the solutions (which has been mentioned above) in the past but unfortunately nothing has helped.

If you get any helpful solution from SAP, please do update us.

Regards,

Neetesh

Ask a Question

Top Q&A Solution Author

User

Count