Hi Bjoern,

Thank you for the response. Thanks also for your work on Docker'izing Galaxy. Using the Tool Shed to create Dockerfiles also appears to have good potential -- thanks for exploring that possibility, Greg.

The crux of my question is on whether it is possible to clone a customized Galaxy installation (whether using Docker containers or VM). 

As an example, the "shed_tool_data_table_conf.xml" file contains absolute paths to the ".loc" files containing reference genomes.  Those absolute paths (unless OK to convert them to relative paths or a different set of absolute paths) would prevent a clone of the reference installation from working properly since each cloned instance would require a unique "galaxy-dist" and supporting set of directories (within a common cluster). To illustrate:

1. "galaxy-dist" on the master instance is located within "/export/galaxy/master/". Installing BWA using the Tool Shed results in a "shed_tool_data_table_conf.xml" file that contains "/export/galaxy/master/galaxy-dist/tool-data/toolshed.g2.bx.psu.edu/repos/devteam/bwa_wrappers/b4427dbb6ced/bwa_index.loc"

2. "galaxy-dist" on a clone of the master instance is located within "/export/galaxy/clone-1/".  In order for this to work, "shed_tool_data_table_conf.xml" would have to reference the ".loc" files either using relative paths or updated absolute paths to reflect the proper location of the ".loc" files.  Can this be done without causing unknown / TBD side effect ?

I am looking for input on whether there might be other similar (surmountable / unsurmountable) considerations, including path dependencies within the database, that would make cloning a reference installation unviable.

Thanks.

> Date: Mon, 7 Jul 2014 23:21:56 +0200
> From: bjoern.gruening@gmail.com
> To: galaxy-dev@lists.bx.psu.edu; srinivas.galaxy@outlook.com
> Subject: Re: [galaxy-dev] Clonable Galaxy installation
>
> Hi Srinivas,
>
> I'm not sure I entirely understood your questions, its to late for me
> today sorry. But maybe you would like to have a look at:
>
> https://github.com/bgruening/docker-recipes
>
> It's basically a docker base image (galaxy) on top you can create
> several specialised images can be created (e.g. galaxy-deeptools). Every
> child images is 'just' a set of toolshed tools.
>
> Also have a look at https://trello.com/c/tm1fPvyq and try the new Test
> ToolShed feature, which will create Dockerfiles for you!
>
> Let us know how that works for you. It's still all work in progress and
> can change or entirely disappear :)
> Feedbacks, Pull-Requests welcome!
>
> Cheers,
> Bjoern
>
> Am 07.07.2014 23:08, schrieb Srinivas .:
> > Hello,
> > I am writing to inquire if there are considerations which would prevent creation of multiple instances of Galaxy by cloning a reference Galaxy installation containing a pre-configured set of tools and related assets. Both the reference and cloned instances would reside in the same cluster, and are each built within dedicated virtual machines . A Docker based approach has yet to be attempted but the same question likely applies.
> > One issue that was observed in attempting this was that the Tool Shed populates the “shed_tool_data_table_conf.xml” file with absolute paths to the “.loc” files for tools installed via the Tool Shed (even though all path related variables are specified using relative paths in the respective configuration files). Might there be other similar dependencies or constraints, including those in the database, that would disallow this clonable Galaxy installation model from being viable ?
> > With regard to the Tool Shed updating files with absolute paths: Could the Toolshed be configured to use relative paths ? Else, would it be OK to update the "file path" variables within the instances created from the reference installation to reflect the instance-specific file paths ?
> > Thanks in advance.
> > --Srinivas
> >
> >
> >
> >
> > ___________________________________________________________
> > 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:
> > http://lists.bx.psu.edu/
> >
> > To search Galaxy mailing lists use the unified search at:
> > http://galaxyproject.org/search/mailinglists/
> >