Hi team,
After upgrade to 6.2 (5.13 -> 5.14 -> 6.0 -> 6.1 -> 6.2), we have noticed that contents of 'uploads' directory are not clearing automatically. On Production I had to setup a cronjob to manually clear it periodically.
Screenshot-2024-06-01-at-3.38.53PM.png
On demo site, still seeing files from May 9 after upgrade
Interesting. Are you using dvwebloader or other addons? Globus?
Hi Phil,
Only using the Web uploader, we don't have Globus setup for Production yet.
Do you think it could be dvwebloader? @Sherry Lake are you seeing the same thing? Files sticking around?
@Bikram I'm asking in IQSS Slack if anyone has any ideas.
For us its happening on all Test, Demo, Prod environments
Philip Durbin said:
Do you think it could be dvwebloader? Sherry Lake are you seeing the same thing? Files sticking around?
We haven't upgraded to 6.2 yet. We are doing that (from 5.14) next week.
Thanks, Sherry.
@Bikram honestly, if you want to go ahead and create an issue, please feel free. Where there's smoke, there's fire, as we say. :grimacing:
Thanks Phil, will create issue.
This looks great, thanks: Temporary files in uploads directory not deleting automatically #10667
I can confirm this. Is it safe to remove those .tmp files manually?
AFAIK once the file is uploaded and user hit the save button, those .tmp files are of no use and should be deleted automatically. I have added a cronjob to run every 2hours and deleted files older than 2hrs. It should give users enough time to save the changes.
hey @Bikram what type of datastore were you testing? my GRA was unable to reproduce this in version 6.3 with a local datastore. next to set up an S3 bucket.
Hi @Don Sizemore, we are using S3 datastore but Openstack Swift S3.
We are using 6.6 and experiencing this issue. Small files are left behind (assuming success?) and if something fails, large ones get stuck there.
Looks like this was fixed in 6.4, but I'm wondering if we missed doing something that would have helped?
-rw-r--r--. 1 dataverse dataverse 11 May 28 03:52 upload_5df8c250_19e698dd064__7feb_00000125.tmp
-rw-r--r--. 1 dataverse dataverse 11 May 28 03:52 upload_5df8c250_19e698dd064__7feb_00000124.tmp
-rw-r--r--. 1 dataverse dataverse 47 May 28 03:52 upload_5df8c250_19e698dd064__7feb_00000137.tmp
-rw-r--r--. 1 dataverse dataverse 22 May 28 03:52 upload_5df8c250_19e698dd064__7feb_00000136.tmp
-rw-r--r--. 1 dataverse dataverse 45 May 28 03:52 upload_5df8c250_19e698dd064__7feb_00000135.tmp
-rw-r--r--. 1 dataverse dataverse 4 May 28 03:52 upload_5df8c250_19e698dd064__7feb_00000134.tmp
-rw-r--r--. 1 dataverse dataverse 39 May 28 03:52 upload_5df8c250_19e698dd064__7feb_00000133.tmp
-rw-r--r--. 1 dataverse dataverse 5 May 28 03:52 upload_5df8c250_19e698dd064__7feb_00000131.tmp
-rw-r--r--. 1 dataverse dataverse 11 May 28 12:55 upload_5df8c250_19e698dd064__7feb_00000139.tmp
-rw-r--r--. 1 dataverse dataverse 47 May 28 12:55 upload_5df8c250_19e698dd064__7feb_00000148.tmp
We will add something to crontab - these don't indicate any current processing is happening? I just want to make sure we set a time that's long enough that we don't hurt current uploads. I see from the bug ticket that IQSS clears out files older than 4 hours, every 30 mins.
Our dataverse.files.directory looks similar with larger files -- is stuff sitting there for ingest? When I can I assume it's safe to delete it?
Thanks!
@Bethany Seeger huh, it's strange that you're on 6.6 and still seeing this issue. Like you said a fix (#10867) went into 6.4.
@Bethany Seeger we're still seeing these on v6.8; we dealt with it by setting <jvm-options>-Ddataverse.files.uploads=/tmp</jvm-options> so they get written to /tmp and (on a Red Hat system) the tmpdir-cleaner systemd service will nuke them after 10 days.
Odd that you're still seeing it too, @Don Sizemore . We are going to setup some cron jobs (we had them, but then a path changed).
Should we re-open the ticket?
@Bethany Seeger I'd prefer a fresh issue, if you don't mind creating one!
Last updated: May 30 2026 at 09:11 UTC