Skip to content

GH-4396: remove workaround for full disk - #4397

Merged
OyvindLGjesdal merged 1 commit into
apache:mainfrom
OyvindLGjesdal:GH-4396-windows-gh-action-disk-full
Nov 12, 2026
Merged

OyvindLGjesdal merged 1 commit into
apache:mainfrom
OyvindLGjesdal:GH-4396-windows-gh-action-disk-full

Conversation

@OyvindLGjesdal

@OyvindLGjesdal OyvindLGjesdal commented Nov 10, 2026 •

Copy link
Copy Markdown
Contributor

GitHub issue resolved #4396

Pull request Description:

Passes on https://github.com/OyvindLGjesdal/jena/actions/runs/42920255479, did not find branch to run on apache/jena.


  • Tests are included.
  • Documentation change and updates are provided for the Apache Jena website
  • Commits have been squashed to remove intermediate development commit messages.
  • Key commit messages start with the issue number (GH-xxxx)

By submitting this pull request, I acknowledge that I am making a contribution to the Apache Software Foundation under the terms and conditions of the Contributor's Agreement.


See the Apache Jena "Contributing" guide.

@OyvindLGjesdal OyvindLGjesdal changed the title GH-4396: remove workaround for disk full GH-4396: remove workaround for full disk Nov 10, 2026
@afs

afs commented Nov 11, 2026

Copy link
Copy Markdown
Member

This will help - I think it frees up about 1.2G.

There is also the problem of all the memory mapped files that, on Windows, don't get freed until the whole surefire part finishes.

I tried without surefire in TDB1 and the build ran on Windows.

So if we hit the limit again, reducing the TDB1 uses of disk is another step. I'm sure many tests could use in-memory TDB1 and we then leave a small number that check on-disk. TDB2 covers the operating system calls and has very similar code to TDB1.

@rvesse

rvesse commented Nov 11, 2026 •

Copy link
Copy Markdown
Member

There is also the problem of all the memory mapped files that, on Windows, don't get freed until the whole surefire part finishes.

We could also play with the Surefire <forkCount> with <reuseForks> as false in the configuration which would cause Surefire to split tests across multiple isolated forks potentially allowing Windows to free memory mapped files sooner?

@OyvindLGjesdal

OyvindLGjesdal commented Nov 12, 2026 •

Copy link
Copy Markdown
Contributor Author

It looks like storage has been moved around: so we should be OK as long as we are on the default disk D:
temp and jena on D: are slightly above 31 GB, while D: has 117 GB free.

Filesystem            Size  Used Avail Use% Mounted on
C:/Program Files/Git  150G  120G   30G  80% /
D:                    150G   34G  117G  23% /d
Largest directories:
560M	/d/a/_temp/m2/repository
1.2G	/d/a/_temp
30G	/d/a/jena/jena
2.0K	/c/Users/runneradmin/.m2
331M	/c/Users/runneradmin/AppData/Local/Temp
328M	/c/Users/runneradmin/AppData/Local/Yarn
851M	/c/Users/runneradmin/AppData/Local/Cypress
Largest target directories:
3.1M	./jena-base/target
6.8M	./jena-ontapi/target
8.2M	./jena-geosparql/target
14M	./jena-fuseki2/jena-fuseki-ui/target
23M	./jena-core/target
42M	./jena-arq/target
45M	./jena-fuseki2/jena-fuseki-geosparql/target
47M	./apache-jena/target
55M	./jena-fuseki2/jena-fuseki-server/target
82M	./jena-db/jena-dboe-base/target
98M	./jena-fuseki2/apache-jena-fuseki/target
581M	./jena-integration-tests/target
1.4G	./jena-fuseki2/jena-fuseki-main/target
3.6G	./jena-tdb2/target
23G	./jena-tdb1/target

@OyvindLGjesdal
OyvindLGjesdal merged commit c32d2a8 into apache:main Nov 12, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Disk full for windows GH action

3 participants