I have implemented a basic script to perform administrative dataset
deletion and email notification. Right now it's limited to the
history_dataset_assocation update_time and an optional tool_id string.
I have pushed it to a galaxy-central fork and will issue a pull request
once I've tested it a bit more. I'm open to comments or suggestions, it
could certainly be extended. Hopefully people find this useful.
Mark datasets as deleted that are older than specified cutoff
and (optionaly) with a tool_id that matches the specified search
This script is useful for administrators to cleanup after users who
leave many old datasets around. It was modeled after the
script originally distributed with Galaxy.
admin_cleanup_datasets.py universe_wsgi.ini -d 60 \
config_file - the Galaxy configuration file (universe_wsgi.ini)
-d --days - number of days old the dataset must be (default: 60)
--tool_id - string to search for in dataset tool_id
--template - Mako template file to use for email notification
-i --info_only - Print results, but don't email or delete anything
-e --email_only - Email notifications, but don't delete anything
Useful for notifying users of pending deletion
--smtp - Specify smtp server
If not specified, use smtp settings specified in config file
--fromaddr - Specify from address
If not specified, use error_email_to specified in config file
Email Template Variables:
cutoff - the cutoff in days
email - the users email address
datasets - a list of tuples containing 'dataset' and 'history' names
I have same questions as yours about cleanup data in Galaxy.
We maintain a local instance of the Galaxy system. I am thinking a way to
delete datasets which are not accessed/updated for a certain period of time, no
matter if users deleted them. In addition, sending an email to users before
deleting their dataset. It looks like the current scripts for clean up only
purges deleted histories/datasets. I was trying to find a way to use API
functions to delete old files which are not deleted, but am not successful. I
found the update_time is not access time in the dataset table. A reference file
may be used frequently but the update_time is pretty old. This would be a
problem if deleting file by the update_time. I think files should be deleted by
the access time. Is it enough to resolve this problem by checking the
update_time of the HistoryDatasetAssociation table?
Thanks for your help,
Joint Genome Institute, LBNL
<quote author='Lance Parsons'>
Nate Coraor wrote:
On Mar 22, 2013, at 11:56 AM, Lance Parsons wrote:
I have been running a Galaxy server for our sequencing researchers for a
while now and it's become increasingly successful. The biggest resource
challenge for us has been, and continues to be disk space. As such, I'd
like to implement some additional cleanup scripts. I thought I run a few
questions by this list before I got too far into things.
In general, I'm wondering how to implement updates/additions to the
cleanup system that will be in line with the direction that the Galaxy
project is headed. The pgcleanup.py script is the newest piece of code
in this area (and even adds cleanup of exported histories, which are
absent from the older cleanup scripts). Also, the pgcleanup.py script
uses a "cleanup_event" table that I don't believe is used by the older
cleanup_datasets.py script. However, the new pgcleanup.py script only
works for Postgres, and worse, only for version 9.1+. I run my system on
RedHat (CentOS) and thus we use version 8.4 of Postgres. Are there plans
to support other databases or older versions of Postgres?
pgcleanup.py makes extensive use of Writable CTEs, so there is not really
a way to port it to older versions. For 8.4 or MySQL, you can still use
the older cleanup_datasets.py.
After looking at it a bit more, I see what you mean. Are there plans to
implement and additional cleanup scripts for non-postgres 9.1 users?
Just curious so I don't reinvent the wheel, I'd be happy to help with
I'd like to implement a script to delete (set the deleted flag) for
certain datasets (e.g. raw data imported from our archive, for old,
inactive users, etc.). I'm wondering if it would make sense to try and
extend pgcleanup.py or cleanup_datasets.py. Or perhaps it would be best
to just implement a separate script, though that seems like I'd have to
re-implement a lot of boilerplate code for configuration reading,
connections, logging, etc. Any tips on generally acceptable (supported)
procedures for marking a dataset as deleted?
You could probably reuse a lot of the code from either of the cleanup
scripts for this.
Right. It seems to make sense to me to focus on the cleanup_datasets.py
since that will work for everyone. I would like to essentially mimic
the user deleting a dataset. I'd then email them to let them know that
some old data had been marked for deletion and let the rest of the
scripts proceed as normal, cleaning that up if they don't undelete it.
It looks like I would want to mark the HistoryDatasetAssociations as
deleted? Is that correct? Would I need to do anything else to simulate
the user deleting the dataset?
Thanks for the help,
Lance Parsons - Scientific Programmer
134 Carl C. Icahn Laboratory
Lewis-Sigler Institute for Integrative Genomics
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: