[
https://issues.apache.org/jira/browse/PDFBOX-2688?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Tilman Hausherr updated PDFBOX-2688:
------------------------------------
Description:
I am running with 6GB of heap space and running PDF to PNG conversion in a
servlet container(tomcat). This happens only when running thousands of requests
for conversion.
JVM memory statistics shows heap space never going above 1GB and non-heap
memory is also constant but the linux process or windows process seems to
consume around 8GB of memory.
Heap dump shows that the largest object is sun.java2d.Disposer and is around
200MB.
It seems that the leaked memory is native memory used by java2d and not
accounted in the heap memory statistic but this growth of sun.java2d.Disposer
memory is proportional to the growth of process memory(linux 'top' command).
{code}
BufferedImage image = null;
ByteArrayInputStream pdfStream = getpdfbytesfromExistingDoc();
PDDocument document = null;
PDPage page = null;
COSDocument cosDoc = null;
PDFParser parser = null;
try {
parser = new PDFParser(pdfStream);
parser.parse();
cosDoc = parser.getDocument();
document = new PDDocument(cosDoc);
@SuppressWarnings("unchecked")
List<PDPage> pages =
document.getDocumentCatalog().getAllPages();
page = pages.get(0);
int imageType = BufferedImage.TYPE_INT_ARGB;
image = page.convertToImage(imageType, 72);
} finally {
if (cosDoc != null) {
cosDoc.close();
}
if (parser != null) {
parser.clearResources();
}
if (document != null) {
if (page != null) {
page.clear();
}
document.close();
}
}
return image;
}
{code}
was:
I am running with 6GB of heap space and running PDF to PNG conversion in a
servlet container(tomcat). This happens only when running thousands of requests
for conversion.
JVM memory statistics shows heap space never going above 1GB and non-heap
memory is also constant but the linux process or windows process seems to
consume around 8GB of memory.
Heap dump shows that the largest object is sun.java2d.Disposer and is around
200MB.
It seems that the leaked memory is native memory used by java2d and not
accounted in the heap memory statistic but this growth of sun.java2d.Disposer
memory is proportional to the growth of process memory(linux 'top' command).
BufferedImage image = null;
ByteArrayInputStream pdfStream = getpdfbytesfromExistingDoc();
PDDocument document = null;
PDPage page = null;
COSDocument cosDoc = null;
PDFParser parser = null;
try {
parser = new PDFParser(pdfStream);
parser.parse();
cosDoc = parser.getDocument();
document = new PDDocument(cosDoc);
@SuppressWarnings("unchecked")
List<PDPage> pages =
document.getDocumentCatalog().getAllPages();
page = pages.get(0);
int imageType = BufferedImage.TYPE_INT_ARGB;
image = page.convertToImage(imageType, 72);
} finally {
if (cosDoc != null) {
cosDoc.close();
}
if (parser != null) {
parser.clearResources();
}
if (document != null) {
if (page != null) {
page.clear();
}
document.close();
}
}
return image;
}
> sun.java2d.Disposer leak when using pdf to image conversion in a
> server(tomcat)
> -------------------------------------------------------------------------------
>
> Key: PDFBOX-2688
> URL: https://issues.apache.org/jira/browse/PDFBOX-2688
> Project: PDFBox
> Issue Type: Bug
> Affects Versions: 1.8.8
> Reporter: Ankit Khanal
>
> I am running with 6GB of heap space and running PDF to PNG conversion in a
> servlet container(tomcat). This happens only when running thousands of
> requests for conversion.
> JVM memory statistics shows heap space never going above 1GB and non-heap
> memory is also constant but the linux process or windows process seems to
> consume around 8GB of memory.
> Heap dump shows that the largest object is sun.java2d.Disposer and is around
> 200MB.
> It seems that the leaked memory is native memory used by java2d and not
> accounted in the heap memory statistic but this growth of sun.java2d.Disposer
> memory is proportional to the growth of process memory(linux 'top' command).
> {code}
> BufferedImage image = null;
> ByteArrayInputStream pdfStream = getpdfbytesfromExistingDoc();
> PDDocument document = null;
> PDPage page = null;
> COSDocument cosDoc = null;
> PDFParser parser = null;
> try {
> parser = new PDFParser(pdfStream);
> parser.parse();
> cosDoc = parser.getDocument();
> document = new PDDocument(cosDoc);
> @SuppressWarnings("unchecked")
> List<PDPage> pages =
> document.getDocumentCatalog().getAllPages();
> page = pages.get(0);
> int imageType = BufferedImage.TYPE_INT_ARGB;
> image = page.convertToImage(imageType, 72);
> } finally {
> if (cosDoc != null) {
> cosDoc.close();
> }
> if (parser != null) {
> parser.clearResources();
> }
> if (document != null) {
> if (page != null) {
> page.clear();
> }
> document.close();
> }
> }
> return image;
> }
> {code}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscr...@pdfbox.apache.org
For additional commands, e-mail: dev-h...@pdfbox.apache.org
