Re: 7.0.1 leak?

2016-08-25 Thread Felipe Jaekel
Thanks for the tips, I'm now able to use visualvm remotely. I inspected some head dumps with MAT and found two leaks related to the Amazon SDK... 2016-07-22 12:13 GMT-03:00 Steve Goldsmith : > Oh and -Dcom.sun.management.jmxremote.authenticate=false, I authenticate > with mine, so I missed t

Re: 7.0.1 leak?

2016-07-22 Thread Steve Goldsmith
Oh and -Dcom.sun.management.jmxremote.authenticate=false, I authenticate with mine, so I missed this for yours. On Fri, Jul 22, 2016 at 11:07 AM, Steve Goldsmith wrote: > You can use one port by doing the following (no random port to mess with > firewall): > > TOMEE/conf/server.xml > > classNam

Re: 7.0.1 leak?

2016-07-22 Thread Steve Goldsmith
You can use one port by doing the following (no random port to mess with firewall): TOMEE/conf/server.xml Download catalina-jmx-remote.jar for version of Tomcat you are running (You'll see version when you go to root page in TomEE or catalina.out). You can download from http://tomcat.apache.org

Re: 7.0.1 leak?

2016-07-22 Thread Romain Manni-Bucau
default remote jmx system needs 2 ports and one is random per default IIRC, jmxmp protocol needs one more jar but avoids this issue ( https://rmannibucau.wordpress.com/2014/04/22/tomee-makes-jmxmp-usage-simple/ ) Romain Manni-Bucau @rmannibucau | Blog

Re: 7.0.1 leak?

2016-07-22 Thread Felipe Jaekel
Hi, I added the the heap dump args, next time JVM crashes I'll try MAT. I ran VisualVM with 7.0.1 and 7.0.2 locally 7.0.1 had 1655 org.apache.openejb.Logger instantes (39720 bytes) 7.0.2 had 54 org.apache.openejb.Logger instantes (1296 bytes) I'd like to run VisualVM remotely, but I was not able

Re: 7.0.1 leak?

2016-07-21 Thread Steve Goldsmith
You need to set up JMX and use jvisualvm to grab a heap dump. Or add JVM args -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath to have one created when you run out of memory next time.​ Analyzing the heap dump is the only true way to find the leak. Also, have a look at MAT http://www.eclipse.org/ma

Re: 7.0.1 leak?

2016-07-21 Thread Romain Manni-Bucau
Until you know what the leak is (jvisualvm can help if you get a heap dump, it can be done through JMX IIRC) it is hard to help you. If you find the thread on the forum/mails where we treated it with Steve you will see what it should look like if you are affected by this bug. If you have several E

Re: 7.0.1 leak?

2016-07-21 Thread Felipe Jaekel
I changed a lot of lib versions when upgrading to 7.0.1, but I guess the most relevant in this case is Hibernate, from 4.2.21 to 5.2.1. I'll try 7.0.2-SNAPSHOT, but I'd like to check locally first if it is going to leak. Is there a simple way to do that? 2016-07-21 8:52 GMT-03:00 Romain Manni-Bu

Re: 7.0.1 leak?

2016-07-21 Thread Romain Manni-Bucau
If you provide slf4j-api without an implementation in the webapp you likely use slf4j-api (and impl) from the container. Identified leak was about openejb loggers and is not really linked to the logger framework used. I didn't check the JSF case to be honest but JMS and failing deployment was 2 id

Re: 7.0.1 leak?

2016-07-21 Thread Felipe Jaekel
One of my webapps has slf4j-api-1.7.21.jar, the others have slf4j-api-1.7.7.jar. No slf4j implementation and no logback. I use a lot of Logger.getLogger... I also have a lot of MyFaces warning logs, but I don't know how MyFaces logs. Is this enough to leak or something else is causing the proble

Re: 7.0.1 leak?

2016-07-20 Thread Steve Goldsmith
Yea, it took a couple weeks for me to burn through 2G. I have one app that logs 1M+ lines a day, so that helped accelerate it. Now TomEE only consumes ~200M with the logger fix applied. On Wed, Jul 20, 2016 at 4:56 PM, Romain Manni-Bucau wrote: > was leaking with JUL as well but you should get a

Re: 7.0.1 leak?

2016-07-20 Thread Romain Manni-Bucau
was leaking with JUL as well but you should get a bunch of errors in logs Romain Manni-Bucau @rmannibucau | Blog | Old Wordpress Blog | Github | LinkedIn <

Re: 7.0.1 leak?

2016-07-20 Thread Steve Goldsmith
This had to do with using logback and slf4j in my WAR. It was fixed in 7.0.2 snapshot. On Wed, Jul 20, 2016 at 4:16 PM, Felipe Jaekel wrote: > Hi, > > After upgrading from 1.7.2 to 7.0.1, server is crashing after about 10 > hours running: > > OpenJDK 64-Bit Server VM warning: INFO: > os::commit_

7.0.1 leak?

2016-07-20 Thread Felipe Jaekel
Hi, After upgrading from 1.7.2 to 7.0.1, server is crashing after about 10 hours running: OpenJDK 64-Bit Server VM warning: INFO: os::commit_memory(0x0007, 1073741824, 0) failed; error='Cannot allocate memory' (errno=12) Found this thread: http://tomee-openejb.979440.n4.nabble.com/To