I usually use SMO 3.1P1, but if you’ve already got that, then there’s one other thing to consider. Some customers have an unusually large number of archive logs and SMO can have problems cataloging them all.

First, locate the launch-java file in /opt/NetApp or /opt/netapp and locate a line that is similar to this:

java -Xmx96m -Djava.net.preferIPv4Stack=true -XX:+HeapDumpOnOutOfMemoryError -enableassertions ${JVM_ARGS} -cp ${CLASSPATH} ${1+"$@"}

The “96”  there is the max heap size. Try changing that to 128 or 256 and see if the problem disappears. If not, then the next step would be to open a support case.

From: Dan Finn [mailto:dan_j_finn@yahoo.com]
Sent: Friday, September 30, 2011 7:30 PM
To: Steiner, Jeffrey
Cc: toasters@teaparty.net
Subject: Re: SMO dumping java hprof files

Thanks Jeffrey,

Do you know how I could go about getting more info on those bugs?  I just checked on the now site and it does seem like we are running the latest version of SMO, 3.1.

Thanks,

Dan

---

From: "Steiner, Jeffrey" <Jeffrey.Steiner@netapp.com>
To: Dan Finn <dan_j_finn@yahoo.com>
Sent: Friday, September 30, 2011 6:27 AM
Subject: RE: SMO dumping java hprof files

There is a bug or two fixed in one of the SMO P releases that can lead to java crashes during operations, and the result is this big hprof files left behind. You definitely should investigate. That only happens when a java process crashes.

From: Dan Finn [mailto:dan_j_finn@yahoo.com]
Sent: Thursday, September 29, 2011 7:09 PM
To: toasters@teaparty.net
Subject: SMO dumping java hprof files

Hello,

We recently noticed that our / partition on one of our oracle RAC servers was filling up.  While investigating I noticed that snap manager for oracle seems to be dumping files under /opt/Netapp/smo named java_pidXXXX.hprof.  These files are all around 90 meg.  It appears that this has something to do with java being run with the "-XX:+HeapDumpOnOutOfMemoryError". 

I'm very new to this environment but from what I'm being told, there is no issues with SMO, it's functioning correctly and as it always has.  I am a little concerned though that these dumps are being generated due to an out of memory error.

Someone did mention that they thought during the initial setup they may have turned on debug logging but they aren't positive and they don't remember how it was turned it on.

Has anyone else ever run into this?

We are running:

SMO 3.1

SnapDrive 4.2

Oracle 11g

RHEL 5.4

Using both NFS and FCP

ASM is mounting the FCP luns

Thanks,

Dan