How to monitor OOM error

[pedro_baroni3]

Hello Experts,

Periodically I'm having problems with some Enqueue-Locks and Communication Channels JDBC Sender stop processing messages.

Perform the procedure in item 29 of the SAP Note "831162 - FAQ: XI 3.0 / 7.0 IP / IP 7.1 JDBC Adapter" Communication Channels and JDBC Sender return to work.

I've checked the message size and the largest is only 1MB.

Also no error in the SQL Query in the database.

How do I track the OOM errors that generate the Enqueue-Locks?

Tks in advance.

[vadimklimov · Accepted solution]

Hello Pedro,

Did you check log in std_server*.out for OOM exceptions?

My regards,

Vadim

[pedro_baroni3 · Accepted solution]

Hi Vadim,

I checked the log files std_server*.out, but there are no exceptions.

There are other places to be checked? There may be other causes for the Enqueue-Lock?

Thanks in advance.

[vadimklimov · Accepted solution]

Hello Pedro,

Speaking about enqueue locks, do JDBC communication channels close sessions to the integrated database correctly? (e.g. is option for disconnecting from database after message processing is enabled in communication channels?)

For more precise analysis, could you generate thread dumps when affected JDBC channels are running and complete their work and check obtained dumps for JDBC channels behaviour regarding database session closure?

My regards,

Vadim

[pedro_baroni3 · Accepted solution]

Vadim,

Yes, the option for disconnecting from database after message processing is enabled in communication channels.

An error in a JDBC Receiver Communication Channel could generate the enqueue-lock in JDBC Sender?

Best Regards.

[pedro_baroni3 · Accepted solution]

Hi Experts,

Any idea?

Tks.

[Former Member · Accepted solution]

Hi Pedro, did you try in NWA ? , there are a lot of monitor to view logs instead of see directly into server file system,

About the problem , try to set de Maximun concurrency=1 parameter in the channel , i had it this problem with a channel JBDC that had a high numer of connections, please let me know if this work for you.

Best regards

[pedro_baroni3 · Accepted solution]

Hi Francisco,

I will try. But right now I'm looking for errors in Procedure in the DB of the legacy system that could be generating the Locks, as I've checked and there are no OOM errors.

Thank you all.

Pedro Baroni

[former_member241146]

Pedro.

See this SAP NOTES:

Note 937159 - XI Adapter Engine is stuck

Note 1083488 - XI FTP/JDBC sender channel stop polling indefinitely(04/04S)

Moreover you can look into Visual Admin >Cluster> "LOCKING ADAPTER" , option of Display Locks then check for an entry with NAME : $XIDBAD.JDBC2XI . Reset the locks and restart the CC.

It's to PI 7.0

Regards.

Bruno

[former_member241146]

Hi Pedro. How are you?

There is option of "Database Locks" in this path:

RWB -> Component monitoring -> Adapter Engine -> Engine Status button (near of CC Monitoring) -> Database Locks

Also you can see Java locks in the NWA tools (only PI 7.1, I guess).

NWA -> Availability and Performance Management ->Resource Monitoring -> LOCKS(system locks java).

I hope it helps you.

Best Regards.

Bruno

[marksmyth]

Hello

Check the note #1289240 How to deal with out of memory errors. Set the JVM properties as documented in the note. When the -XX:+HeapDumpOnOutOfMemoryError parameter is set a Java Heap dump will be automatically created when an OutOfMemoryError occurs.

The heap dump is generated in directory /usr/sap/<SID>/<INSTANCE>/j2ee/cluster/server<N>/ where N is the server node number.

You can then use the Eclipse Memory Analyzer (see section 4 of the note) to analyse the OOM Error.

Regards

Mark