hinerm

2023-03-24 11:22:00
Full thread dump OpenJDK 64-Bit Server VM (25.332-b09 mixed mode):

"RMI TCP Connection(2)-192.168.68.53" #66 daemon prio=5 os_prio=0 tid=0x00000173573c2800 nid=0x55a8 runnable [0x000000648192d000]
   java.lang.Thread.State: RUNNABLE
        at java.net.SocketInputStream.socketRead0(Native Method)
        at java.net.SocketInputStream.socketRead(SocketInputStream.java:116)
        at java.net.SocketInputStream.read(SocketInputStream.java:171)
        at java.net.SocketInputStream.read(SocketInputStream.java:141)

hinerm

#@ OpService ops
#@ ImgPlus image
#@OUTPUT ImgPlus copy
#@OUTPUT Integer a

a = 21
copy = ops.copy().img(image.getImg())
c = copy.cursor()
c.fwd()
v = c.get().copy()

hinerm

package test;

import java.util.StringJoiner;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.TimeUnit;
import java.util.regex.Matcher;
import java.util.regex.Pattern;

public class Sandbox {

hinerm

package test;

import java.util.StringJoiner;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.TimeUnit;
import java.util.regex.Matcher;
import java.util.regex.Pattern;

public class Sandbox {

hinerm

https://docs.google.com/presentation/d/1E_sYCh00fSGYmSue_PRifqaGJ61RnstMdcX5c_c1jWM/edit?usp=sharing

hinerm

The following images show memory leaks due to mapped images not being cleaned up.

The MemoryOverview shows two hard reference paths keeping byte[] instances alive (blue and green). They both go through python classes to a HashMap.

The Objects1 and Objects2 images show these two paths expanded and reveals that we likely have two paths to the same map, and thus to the same set of objects.

Persumably our Python code is making these mappings but not cleaning up after them. Persumably there is also a hard reference to these images in ImageJ itself, so it would make sense on the Python side to cache WeakReferences to the images - or add some callback that ensures these mappings are eventually removed (or manually clean the maps at the end of the script).

To find the problematic python code we can look for the relevant variable names (snapshotPixels, rLUT1, bLUT1, gLUT1, etc...)

hinerm

// search for TODO sections to see parameters that still need tuning

//This macro takes a 3-channel 16 bit immunofluorescance image as input.
//  channel 1 (red): SC
//  channel 2 (green): foci
// channel 3 (blue): centromeres

//The primary goal of this macro is to identify rois covering main cell features,
//measure the total distance of skeletonized SC, number of foci on each SC and distance from foci to the centromere on the same skeleton
// Requirements:

hinerm

package org.scijava;
import java.io.FileOutputStream;
import java.io.IOException;

public class Main {

     public static void main(String... args) throws IOException {

        //FIXME: set this path to the location of the jar containing this main class.
        String url = "C:/Users/Johannes/loci/test/target/scijava-tests-2.43.1-SNAPSHOT.jar";

hinerm

import java.io.BufferedReader;
import java.io.IOException;
import java.io.InputStreamReader;
import java.lang.reflect.InvocationTargetException;
import java.lang.reflect.Method;
import java.net.URL;
import java.net.URLClassLoader;

public class Main {

hinerm

<?xml version="1.0" encoding="UTF-8"?>
<?JvmMonitor version="3.8.1"?>
<cpu-profile date="2015/01/25 12:09:02" runtime="8608@localhost" mainClass="org.eclipse.jdt.internal.junit.runner.RemoteTestRunner" arguments="-Dscijava.log.level=error -Dfile.encoding=UTF-8">
	<thread name="main">
		<frame name="io.scif.services.SCIFIODatasetService.create(net.imagej.ImgPlus)" cnt="1" time="0">
			<frame name="io.scif.services.SCIFIODatasetService.datasetService()" cnt="1" time="0"/>
		</frame>
		<frame name="io.scif.services.DefaultDatasetIOService.open(java.lang.String)" cnt="7676" time="175061">
			<frame name="io.scif.config.SCIFIOConfig.imgOpenerSetIndex(int)" cnt="7676" time="0">
				<frame name="io.scif.config.SCIFIOConfig.imgOpenerSetRange(io.scif.img.Range)" cnt="7676" time="0"/>