Besides a couple of small other changes, there are two new features in this release of Mumbai.
First, you can now analyze Oracle heap dump trace files in Mumbai. Once again, this functionality was inspired by a blog post by Tanel Põder. He created a shell script for aggregating the chunk sizes found in heap dump trace files and while I used this script quite a few times, I found it a bit awkward to have it available on
LinuxUnix systems only and not on Windows.
What you can now do is to open a trace file with heap dump information and Mumbai will analyze the trace file and aggregate over the heap names and allocation reason as shown in the screenshot below.
Mumbai will warn you if it finds information about the same heap block (identified by the heap name) more than once, e.g. because the information was dumped to the same trace file twice and you can filter on the last column “Heap block occurrence” to see the aggregated data for the occurrence you wish to see (you might want to filter on the “Trace file line number” column, too).
The second new feature is a dialog to create Oracle memory dump trace files as seen below.
When you check the “Retrieve trace file after execution” box (and the Mumbai42 schema which holds this functionality is available in the database), Mumbai will retrieve the generated trace file and open it and in the case of a memory heap dump trace, you can send it to the heap dump analyzer with another click of a button.
As with lots of functionality in Mumbai, handle with care and make sure you know what you do. Trace files can get quite large and depending on what kind if trace dump you initiate it can bring your database instance to a full stop for some time.