Problems installing conda dependencies for seq2hla tool
Dear Devs I've been having problems installing the jjohnson/seq2hla tool from the toolshed on our production instance (running release_17.09). Specifically Galaxy has been unable to successfully install the seq2HLA 2.2 dependency via conda. However it seems to work fine when installing into a clean local instance. Initially the problem seemed to be that our version of conda was too old (we had 4.2.13), so while the __seq2HLA@2.2 env was being created, the 'seq2hla' package couldn't be installed into it (so Galaxy thought that the requirement was satifsfied but essentially it was broken). I updated conda to 4.3.24 and checked that our "ensure_conda_channels" setting matched the one in 'galaxy.ini.sample'. Now when I try to reinstall the seq2hla tool, the dependency cannot be installed at all - there seems to be an issue with the R 3.1.2-2 conda package which this needs to pull in: ERROR conda.core.link:_execute_actions(337): An error occurred while installing package 'defaults::r-3.1.2-2'. PaddingError: Placeholder of length '80' too short in package /XXXXXXXXX/galaxy/production/tool_dependencies/_conda/envs/__seq2HLA@2.2/bin/Rscript. The package must be rebuilt with conda-build > 2.0. XXXX... above is the path to the Galaxy installation directory, and the full path does exceed 80 characters. Presumably this worked okay on the local Galaxy install because there the path is shorter than the 80 character limit. Is there any workaround or fix that I can use to get this working on our production instance? Also, is it generally safe to update the Galaxy conda installation as I have done? Is there a recommended conda version for Galaxy? Thanks for your help, best wishes Peter -- Peter Briggs peter.briggs@manchester.ac.uk Bioinformatics Core Facility University of Manchester B.1083 Michael Smith Bldg Tel: (0161) 2751482
Just to follow up on my own question, in case it's helpful to anyone else: Re the specific problem with not being able to install the seq2HLA 2.2 dependency via conda: in the end I build my own version and made it available to Galaxy as "Galaxy package" (https://docs.galaxyproject.org/en/master/admin/dependency_resolvers.html#gal...), which seemed to work okay. Re updating to conda 4.3.24: after the update this gave me problems with at least one dependency, due to the "--offline bug" described here: https://github.com/galaxyproject/galaxy/issues/5187 I applied Norbert Eichner's fix in PR #5233 (https://github.com/galaxyproject/galaxy/pull/5233) and reinstalled the conda dependency via "Manage Tool Dependencies" in the "Admin" interface, which seemed to fix that problem. It's not clear to me yet whether other conda dependencies might need to be reinstalled, or if applying the patch is sufficient. Best wishes Peter -- Peter Briggs peter.briggs@manchester.ac.uk Bioinformatics Core Facility University of Manchester B.1083 Michael Smith Bldg Tel: (0161) 2751482 ________________________________________ From: galaxy-dev [galaxy-dev-bounces@lists.galaxyproject.org] on behalf of Peter Briggs [peter.briggs@manchester.ac.uk] Sent: Friday, January 05, 2018 12:06 PM To: galaxy-dev@lists.galaxyproject.org Subject: [galaxy-dev] Problems installing conda dependencies for seq2hla tool Dear Devs I've been having problems installing the jjohnson/seq2hla tool from the toolshed on our production instance (running release_17.09). Specifically Galaxy has been unable to successfully install the seq2HLA 2.2 dependency via conda. However it seems to work fine when installing into a clean local instance. Initially the problem seemed to be that our version of conda was too old (we had 4.2.13), so while the __seq2HLA@2.2 env was being created, the 'seq2hla' package couldn't be installed into it (so Galaxy thought that the requirement was satifsfied but essentially it was broken). I updated conda to 4.3.24 and checked that our "ensure_conda_channels" setting matched the one in 'galaxy.ini.sample'. Now when I try to reinstall the seq2hla tool, the dependency cannot be installed at all - there seems to be an issue with the R 3.1.2-2 conda package which this needs to pull in: ERROR conda.core.link:_execute_actions(337): An error occurred while installing package 'defaults::r-3.1.2-2'. PaddingError: Placeholder of length '80' too short in package /XXXXXXXXX/galaxy/production/tool_dependencies/_conda/envs/__seq2HLA@2.2/bin/Rscript. The package must be rebuilt with conda-build > 2.0. XXXX... above is the path to the Galaxy installation directory, and the full path does exceed 80 characters. Presumably this worked okay on the local Galaxy install because there the path is shorter than the 80 character limit. Is there any workaround or fix that I can use to get this working on our production instance? Also, is it generally safe to update the Galaxy conda installation as I have done? Is there a recommended conda version for Galaxy? Thanks for your help, best wishes Peter -- Peter Briggs peter.briggs@manchester.ac.uk Bioinformatics Core Facility University of Manchester B.1083 Michael Smith Bldg Tel: (0161) 2751482 ___________________________________________________________ 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: https://lists.galaxyproject.org/ To search Galaxy mailing lists use the unified search at: http://galaxyproject.org/search/
participants (1)
-
Peter Briggs