Remove objects read permission about galaxy-helm HOT 13 OPEN

The fix is against release 24.0 actually so we won't have to wait for the next release.

from galaxy-helm.

True, but the branch won't be tagged so we won't get a Docker image built automatically (except for the 24.0-auto image). According to John D 24.1.1 will be released in a few days so it should be ok to wait until then.

from galaxy-helm.

Hi @ksuderman have you released the 24.1.1 version with the objects folder permission bug fixed?

from galaxy-helm.

Hi @anik, I did not see a notification about the release, but I see galaxy-min:24.1.1 on quay.io. I am testing it now and will report back.

from galaxy-helm.

Our tests are passing and #481 will be merged shortly.

from galaxy-helm.

Alright. This morning I pulled the 24.1.1 that was available and still the permission denied issue persisted in the object folder. So, I was wondering if the issue was resolved or not.

from galaxy-helm.

Is that with Kraken? I will try to reproduce using that.

from galaxy-helm.

Yes, it is with Kraken. I also tried with Diamond database builder and faced the same error of permission denied.

from galaxy-helm.

Hi @ksuderman could you reproduce and resolve the error? Please let me know.

from galaxy-helm.

Yes I am able to reproduce, at least I think it is the same problem you are experiencing. To be sure though, it is the data_manager_build_kraken2_database tool that is the problem and not kraken2 itself?

I think this is a separate, but related, bug. I am trying to get a few more data points before reaching out to the backend team to find a solution.

However, is there a reason you are trying to build the Kraken2 database manually rather than using the ones available on CVMFS? I see you have refdata.enabled=false in your values.yaml, at least the values shared on the GalaxyHelp channel. It looks like the database you want to use is available as part of the default reference data included on CVMFS.

refdata:
  enabled: true
  type: cvmfs
cvmfs:
  deploy: true  # unless you are deploying CVMFS separately from Galaxy

from galaxy-helm.

Hi @ksuderman thanks for reproducing the error. Yes, the problem is with the kraken2 database builder tool. Kraken2 itself works fine (or would work fine if there is a database).
Regarding your second question, I need to build the database using the Prebuilt RefSeq indexes which are available and getting updated once or twice a year. Also, there is an option of building a custom database with reference genomes other than NCBI. I am not using that at the moment. I am just trying to build it from the available ones and getting the error. For the time being, I can try to change the value file and use your suggestion.

from galaxy-helm.

Hi @ksuderman, I was wondering if there is any update on this issue. Thanks.

from galaxy-helm.

Unfortunately not at this time. The problem is in the way Galaxy runs the post job event handler, which runs in the Galaxy process (pod) and not by the tool. We are still trying to determine why that event handler does not seem to have the proper volumes mounted.

from galaxy-helm.