Configuring Attachment Size
You can limit the size of files that can be uploaded and attached in Confluence.
To configure the maximum file size that can be uploaded:
Go to Administration menu , then General Configuration.
- Choose Edit.
Enter the maximum size next to Attachment Maximum Size.
The default is 100 MB.Choose Save.
Related pages:
How attachments are indexed
When a file is uploaded, Confluence will attempt to extract and index its text. This allows people to search for the content of a file, not just the filename. This process is quite memory intensive and can cause out of memory errors when very large files are uploaded. Confluence has a number of safeguards to prevent this happening:
- If the uploaded file is larger than 100 MB, Confluence will not attempt to extract text or index the file contents. Only the filename will be searchable.
- If the uploaded file is one of the following types, Confluence will only extract up to:
- 1 MB of text from Excel (.xlsx) and PowerPoint (.pptx)
- 8 MB of text from PDF (.pdf)
- 10 MB of text from other text files (including .txt, .xml, .html, .rtf etc)
- 16 MB of text from Word (.docx)
Note that this is based on the size of the file when it's uncompressed. As .xlsx and .docx files are compressed, text extraction may fail even though the size of the file appears to be under the limit.
- If the text extracted from the file was greater than 1 MB, it will be searchable, but Confluence will not show this text as an excerpt with the search result.
If Confluence stops extracting text, only a portion of the file's content will be searchable.
Confluence will only attempt to extract and index the file once. If it fails, it will not try again.
Some of the values above are configurable via system properties. If you experience out of memory errors when people upload large files, you may want to reduce these limits further, using the following properties:
atlassian.indexing.attachment.maxsize
officeconnector.excel.extractor.maxlength
officeconnector.textextract.word.docxmaxsize
atlassian.indexing.contentbody.maxsize
officeconnector.powerpoint.extractor.maxlength