I have now been able to successfully remove datasets from disk. After deleting 
the dataset or history from the front-end interface (as the user), I then run 
the cleanup scripts as admin:

python ./scripts/cleanup_datasets/cleanup_datasets.py ./universe_wsgi.ini -d 0 
-1 $@ >> ./scripts/cleanup_datasets/delete_userless_histories.log
python ./scripts/cleanup_datasets/cleanup_datasets.py ./universe_wsgi.ini -d 0 
-2 -r $@ >> ./scripts/cleanup_datasets/purge_histories.log
python ./scripts/cleanup_datasets/cleanup_datasets.py ./universe_wsgi.ini -d 0 
-3 -r $@ >> ./scripts/cleanup_datasets/purge_datasets.log
python ./scripts/cleanup_datasets/cleanup_datasets.py ./universe_wsgi.ini -d 0 
-5 -r $@ >> ./scripts/cleanup_datasets/purge_folders.log
python ./scripts/cleanup_datasets/cleanup_datasets.py ./universe_wsgi.ini -d 0 
-4 -r $@ >> ./scripts/cleanup_datasets/purge_libraries.log
python ./scripts/cleanup_datasets/cleanup_datasets.py ./universe_wsgi.ini -d 0 
-6 -r $@ >> ./scripts/cleanup_datasets/delete_datasets.log

However, my final goal is to have a process that can remove old datasets from 
disk regardless of whether or not the users have deleted them at the front-end 
(and then automate said process via cronjob). This will be essentially in a 
situation where users are likely to leave datasets unattended and accumulating 
disk space.

I found the following Galaxy thread:


And am trying to use the script it mentions:

python ./scripts/cleanup_datasets/admin_cleanup_datasets.py universe_wsgi.ini 
-d 30 --smtp <smtp server> --fromaddr rsa...@jcvi.org

I chose –d 30 to remove all datasets older than 30 days, which currently only 
targets one dataset. The resulting stdout indicates success:

# 2014-03-25 16:27:47 - Handling stuff older than 30 days
Marked HistoryDatasetAssociation id 301 as deleted

From: rsa...@jcvi.org
To: isi...@jcvi.org
Subject: Galaxy Server Cleanup - 1 datasets DELETED
Galaxy Server Cleanup
The following datasets you own on Galaxy are older than 30 days and have been 

    "Small.fastq" in history "Unnamed history"

You may be able to undelete them by logging into Galaxy, navigating to the 
appropriate history, selecting "Include Deleted Datasets" from the history 
options menu, and clicking on the link to undelete each dataset that you want 
to keep.  You can then download the datasets.  Thank you for your understanding 
and cooporation in this necessary cleanup in order to keep the Galaxy resource 
available.  Please don't hesitate to contact us if you have any questions.

 -- Galaxy Administrators

Marked 1 dataset instances as deleted

But when I check the database, the status of dataset 301 is unchanged 

I then run the same cleanup_datasets.py routine from above (but with –d 30), 
but dataset 301 is still present. I tried a second time, this time using –d 0, 
but still no deletion (which is not surprising since the dataset's deleted 
status is still false).

If I run admin_cleanup_datasets.py again with the same parameters, the stdout 
says no datasets matched the criteria, so it seems to remember it's previous 
execution, but it's NOT actually updating the database.

What am I doing wrong?

Ravi Sanka
ICS – Sr. Bioinformatics Engineer
J. Craig Venter Institute

From: Carl Eberhard <carlfeberh...@gmail.com<mailto:carlfeberh...@gmail.com>>
Date: Tuesday, March 18, 2014 2:09 PM
To: Peter Cock <p.j.a.c...@googlemail.com<mailto:p.j.a.c...@googlemail.com>>
Cc: Ravi Sanka <rsa...@jcvi.org<mailto:rsa...@jcvi.org>>, 
Subject: [CONTENT] Re: [galaxy-dev] Re: Unable to remove old datasets

The cleanup scripts enforce a sort of "lifetime" for the datasets.

The first time they're run, they may mark a dataset as deleted and also reset 
the update time and you'll have to wait N days for the next stage of the 

The next time they're run, or if a dataset has already been marked as deleted, 
the actual file removal happens and purged is set to true (if it wasn't 

You can manually pass in '-d 0' to force removal of datasets recently marked as 

The purge scripts do not check 'allow_user_dataset_purge', of course.

On Tue, Mar 18, 2014 at 11:50 AM, Carl Eberhard 
<carlfeberh...@gmail.com<mailto:carlfeberh...@gmail.com>> wrote:
I believe it's a (BAD) silent failure mode in the server code.

If I understand correctly, the purge request isn't coughing an error when it 
gets to the 'allow_user_dataset_purge' check and instead is silently marking 
(or re-marking) the datasets as deleted.

I would rather it fail with a 403 error if purge is explicitly requested.

That said, it of course would be better to remove the purge operation based on 
the configuration then to show an error after we've found you can't do the 
operation. The same holds true for the 'permanently remove this dataset' link 
in deleted datasets.

I'll see if I can find out the answer to your question on the cleanup scripts.

On Tue, Mar 18, 2014 at 10:49 AM, Peter Cock 
<p.j.a.c...@googlemail.com<mailto:p.j.a.c...@googlemail.com>> wrote:
On Tue, Mar 18, 2014 at 2:14 PM, Carl Eberhard 
<carlfeberh...@gmail.com<mailto:carlfeberh...@gmail.com>> wrote:
> Thanks, Ravi & Peter
> I've added a card to get the allow_user_dataset_purge options into the
> client and to better show the viable options to the user:
> https://trello.com/c/RCPZ9zMF

Thanks Carl - so this was a user interface bug, showing the user
non-functional permanent delete (purge) options. That's clearer now.

In this situation can the user just 'delete', and wait N days for
the cleanup scripts to actually purge the files and free the space?
(It seems N=10 in scripts/cleanup/purge_*.sh at least, elsewhere
like the underlying Python script the default looks like N=60).



Please keep all replies on the list by using "reply all"
in your mail client.  To manage your subscriptions to this
and other Galaxy lists, please use the interface at:

To search Galaxy mailing lists use the unified search at:

Reply via email to