Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
Dominic Fitzgerald
@djf604
The image id is the same, but does it matter that on the compute node the repository and tag are missing?
Andrew Duncan
@agduncan94
I think so :(
Dominic Fitzgerald
@djf604
Ok I'll see what I can do about populating it
I really appreciate your help here @agduncan94 , I brought cookies into work today if you want one and are willing to travel to Chicago haha
Andrew Duncan
@agduncan94
No problem, hopefully we can figure something out that works for you.
haha I’ll hop on the next flight
Dominic Fitzgerald
@djf604
Yep that was it. Once the repo and tag names were populated in "docker images" then the whole thing runs offline
Andrew Duncan
@agduncan94
Awesome :smile:
I've created an issue ga4gh/dockstore#1854
We will likely add the solution to our documentation site, since I'm sure in the future more people will have a similar use case.
How did you manage to populate the repo and tag names?
Dominic Fitzgerald
@djf604
My coworker did it, let me find out
Dominic Fitzgerald
@djf604
Ok just to re-explain my situation, I was able to run the dockstore container just fine on the head node of our commodity cluster, which has internet access. Of course the point of having a cluster is to run software on compute nodes, which do not have public internet access. So when trying to execute the same command on a compute node it was complaining that it couldn't pull down the docker image
So we saved the docker image with "docker save -o dockstore-cgpwgs_with_tags quay.io/wtsicgp/dockstore-cgpwgs:1.1.4" (executed on the head node to some cluster-globally-accessible directory), then added it to the compute node(s) with "docker load -i /path/to/image.tar" (executed on the compute node)
What we had done before, which caused the lack of entries in the repo and tag column was "docker save -o dockstore-cgpwgs.tar 8f2d9de5ab9d", so just giving the tag id doesn't place nicely with dockstore
I did also have to do the whole "dockstore download" business, but that was an easy step that I'm sure is well documented for you
After that, the command "dockstore tool launch --local-entry Dockstore.cwl --json example.json" worked just fine on our offline compute node
Is that helpful?
Andrew Duncan
@agduncan94
this is perfect, thank you
Dominic Fitzgerald
@djf604
No thank you so much. Your help today opened up a weeks long bottleneck, it's much appreciated!
Keiran Raine
@keiranmraine
The onboarding page is not very easy to access once you have an account anymore unless you know the url. Is this intentional or am I blind?
Andrew Duncan
@agduncan94
There is a link on the homepage called setup account, but yeah we were discussing how we can make it more visible after initial onboarding.
The link external accounts is visible in the accounts page, and same with the change username.
Keiran Raine
@keiranmraine
Ah, it used to be under the account area once logged in. Now I've found the API curl I'm good... although I've noticed that if you specify python3 the cwl version downgrades significantly
Gary Luu
@garyluu
ya, i haven't tested with python3 very much, so didn't update the cwl version as frequent
Keiran Raine
@keiranmraine
$ diff -y requirements_py2.txt requirements_py3.txt 
setuptools==36.5.0                        setuptools==36.5.0
cwl-runner                            cwl-runner
cwltool==1.0.20180403145700                      |    cwltool==1.0.20170828135420
schema-salad==2.7.20180514132321                  |    schema-salad==2.6.20170806163416
avro==1.8.1                              |    avro-cwl==1.8.3
ruamel.yaml==0.14.12                        ruamel.yaml==0.14.12
requests==2.18.4                        requests==2.18.4
Andrew Duncan
@agduncan94
Oh yes so we had it visible up until users perform certain actions which imply they have a decent understanding of Dockstore (such as publish a tool), and then we hide it once they perform any of those actions.
Keiran Raine
@keiranmraine
Ah, okay good to know. We need the 2018 features for monitoring docker containers
Andrew Duncan
@agduncan94
We are going to add a link to the onboarding on the accounts page as well in the next hotfix I believe
Also we will create an issue for testing with python 3
Denis Yuen
@denis-yuen
@keiranmraine do you have a link to the 2018 features that you're referring to?
Keiran Raine
@keiranmraine
Options for recording the Docker container identifier into a file:
  --record-container-id
                        If enabled, store the Docker container ID into a file.
                        See --cidfile-dir to specify the directory.
  --cidfile-dir CIDFILE_DIR
                        Directory for storing the Docker container ID file.
                        The default is the current directory
  --cidfile-prefix CIDFILE_PREFIX
                        Specify a prefix to the container ID filename. Final
                        file name will be followed by a timestamp. The default
                        is no prefix.
Sorry @denis-yuen , we want to be able to point some monitoring tools to the correct containers via the cidfiles
Michael R. Crusoe
@mr-c
@keiranmraine you can use cwltool directly, it speaks GA4GH TRS to dockstore.org https://github.com/common-workflow-language/cwltool#use-with-ga4gh-tool-registry-api
Keiran Raine
@keiranmraine
Hi @mr-c, we are using it directly in many cases. I raised it as we build our base image using the requirements.txt from dockstore (as it's via a REST call) so people can choose how they interact.
(I've raised a small, unrelated, Q on cwl gitter channel too)
Keiran Raine
@keiranmraine
@denis-yuen @agduncan94 cwl-runner complains about the absence of html5lib (py2), would be good to get it into the requirements REST route.
Michael R. Crusoe
@mr-c
@keiranmraine cwl-runner could be any CWL implementation, you probably mean the CWL reference runner cwltool (sorry to be pedantic)
Denis Yuen
@denis-yuen
@keiranmraine thanks, I've logged ga4gh/dockstore#1915 for us to update and test with a version of cwltool that supports the monitoring options
Dominic Fitzgerald
@djf604
I'm executing a command: dockstore tool launch --local-entry Dockstore.cwl --json test_run_L000013.json and as far as I can tell the container itself is happy. The pipeline within the container is throwing errors which I think are being saved to logs on the container-local filesystem. Can I view these logs? Does the container instance die when the pipeline terminates?
Andrew Duncan
@agduncan94
Hey Dominic, unfortunately I think that the container dies once the pipeline terminates. We do however have the concept of extra CWLtool parameters that you can apply when running a tool. I can check to see if they have any options which do not terminate the instance.
https://docs.dockstore.org/docs/publisher-tutorials/advanced-features/#running-cwl-runner-with-extra-tags
If you look here it seems like there are options to not delete the container after the pipeline is run
If you add these parameters to your config file then I think you should be able to view the logs by entering the container
If anything is unclear please let me know
Dominic Fitzgerald
@djf604
It seems that the container still dies
At least, I'm able to get into it while it's running but as soon as it errors I get disconnected
It did give me a more detailed trackback though, so I guess that's good
Obviously I don't expect anyone to know specifics about the container I'm running, but does the error message cwltool.errors.WorkflowException: Did not find output file with glob pattern: '['WGS_*_vs_*.result.tar.gz']' bring to mind any general ideas what could be happening?
The command that I think is being run right before the error is battenberg.pl -o /var/spool/cwl/WGS_2016-2066_vs_2016-2067/battenberg -u /var/spool/cwl/reference_files/battenberg/1000genomesloci -e /var/spool/cwl/reference_files/battenbe rg/impute/impute_info.txt -c /var/spool/cwl/reference_files/battenberg/probloci.txt -r /var/spool/cwl/reference_files/genome.fa.fai -ig /var/spool/cwl/reference_files/battenberg/ignore_contigs.txt -ge XX -tb /var/spool/cwl/tmp/2016-2066.bam -nb /var/spool/cwl/tmp/2016-2067.bam -p allelecount -nl 50 -t 40 >& /var/spool/cwl/timings/WGS_2016-2066_vs_2016-2067.time.alleleCount where I don't see anything like the glob above