Kevin Grigorenko: Eclipse Memory Analyzer Tool
Abstract:
How do you analyze Java OutOfMemoryErrors and perform application and thread heap sizing? This talk covers the open source Eclipse Memory Analyzer Tool (MAT) diving into a bit of theory (retained heap, dominator tree, etc.) as well as some tips and tricks learned from the field such as thread stack frame local analysis, sizing applications by classloader, Object Query Language, comparing dumps to find leaks, SoftReference analysis, writing custom MAT queries, and more.
Speaker Bio:
Kevin Grigorenko is a senior Java software engineer on the IBM WebSphere Application Server SWAT team, which provides worldwide, onsite, and remote supplemental product defect support, particularly in critical production situations. He focuses on problem determination for WebSphere Application Server and related stack products and operating systems, including the IBM and HotSpot JVMs, Linux, AIX, Solaris, Windows, z/OS, HP/UX, and IBMi. He is an open source contributor on the Eclipse Memory Analyzer Tool, Wireshark, Eclipse MicroProfile, and OpenLiberty.
www.sdjug.org/2019-05-21
Пікірлер: 10
Very good talk, helped me analyse a heap dump. Lots of good questions answered. Was also interesting to see that little bit of the code of MAT at the end! Thanks Kevin
Thank you for sharing this video Kevin. Your contribution is great !!! I was able to identify an OutOfMemory Error and identify the cause of the problem after watching your video. . .
Thanks for this video! MAT is amaizing tool and this video helped me a lot in understanding how to use it.
thank you so much Kevin, its really helpful
Thank you Kevin this helped me a lot bless you.
muito bom
I have question: how to handle big dump file , like 20GB ? I want to use MAT GUI, so scp 20GB dump_file from server to my pc , but many developer don't have so big memory machine , then I can't analysze it ... Is there any split strategy ?
@gatlesslolita4009
3 жыл бұрын
Yes parsing huge dump file uses a lot of memory! (a rule of thumb is to have as many RAM as the dump file size) The only strategy I found is to parse it with ParseHeapDump.sh tool on a machine with enough RAM. I often use a dedicated machine for this task. Then you can open the generated parsed files on a developer machine with just 1-2 GB of memory available. BTW this presentation is definitely a must see ! The explanations about retained heap, dominator tree, ... are very useful and well explained !
Are these slides available anywhere?
Squeaky chair is a major distraction. Great lecture otherwise.