Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
    9d0cd7d2
    @9d0cd7d2:matrix.org
    [m]
    Hi all, I'm testing Tower to see the capabilities of the tool and it's truly awesome. One doubt that I've is, is it possible to describe the same detail level for the nextflow CLI the compute environments? I mean, using Tower, I can define some parameters for example to create an Slurm cluster: login hostname, login port, etc. How these parameters could be used on a CLI Nextflow pipeline? Using executors? Thanks in advance!
    Paolo Di Tommaso
    @pditommaso
    not always there's an equivalent setting in Nextflow for all options that are available in Tower compute env
    for example, nextlow does not need to be aware of slurm login host, therefore there isn't such option
    9d0cd7d2
    @9d0cd7d2:matrix.org
    [m]
    thanks for the answer @pditommaso , and how can we approach something similar without using Tower? for example, having a central node with Nextflow to deploy processes or pipelines against remote Slurm clusters? Maybe a wrapper or a custom script?
    21 replies
    9d0cd7d2
    @9d0cd7d2:matrix.org
    [m]
    Many many thanks for your answers again @ewels , and of course, congratule always the great work of the devs on the Nextflow/Tower tools because are truly awesome without doubts
    kaitlinchaung
    @kaitlinchaung

    Hi there, thanks for the great tool.

    I’ve noticed lately that the jobs remain in the blue running state, even though they have been cancelled and are not running anymore. Failed jobs and successfully completed jobs are accurately reflected, however. For reference, I am running all my jobs on SLURM. I was wondering if there is maybe something I am doing wrong on my part or if there is any parameter I should change to fix this? I noticed this happening about a month ago. Thanks!

    Paolo Di Tommaso
    @pditommaso
    Hi, we need more details to troubleshoot the problem. Please open an issue her e
    including the workflow id, and the nextflow logs
    Patrick Hüther
    @phue
    Hi, I spotted a mention of tower pipeline reports in the 21.12.0 changelog :mag:
    Looks amazing in the community showcase, how do we enable it on our instance?
    Paolo Di Tommaso
    @pditommaso
    Hi Patrick, this is still a dev preview. We plan to include it into Q1 release for customers
    1 reply
    Phil Ewels
    @ewels
    oooh, I hadn't seen this yet :star2:
    @pditommaso an "open in new tab" button would be nice :eyes: :sweat_smile:
    Paolo Di Tommaso
    @pditommaso
    you are right!
    juhis
    @juhis

    I’ve been looking into Nextflow and Tower and managed to install and run Tower on a cloud VM using a cloud SQL database. Seems very powerful! Monitoring workflows works fine but I can’t seem to get to the API. The service-info endpoint works:

    $ curl -H "Authorization: Bearer TOKEN” http://0.0.0.0:8000/api/service-info
    {"message":null,"serviceInfo":{"version":"20.06.0","commitId":null,"authTypes":[],"loginPath":"/login"}}

    But any other endpoint gives 401 unauthorized:

    $ curl -H "Authorization: Bearer TOKEN” http://0.0.0.0:8000/api/workflow
    frontend_1  | 172.18.0.1 - - [07/Feb/2022:18:27:18 +0000] "GET /api/workflow HTTP/1.1" 401 5 "-" "curl/7.68.0"

    I’m using the tokens I get from the Your tokens page in Tower. Am I missing something in the API call?

    sjroth
    @sjroth
    Is tower.nf down for anyone else?
    Phil Ewels
    @ewels
    Seems to be loading for me!
    James A. Fellows Yates
    @jfy133

    Can anyone suggest any way to debug why on runs 'Resolved configuration' results in n/a on tower.nf?

    Myself and a colleague have the same settings (nxf version, same workspace organisation etc), by my configuration always resolves to n/a but theirs lists all the parameters

    Paolo Di Tommaso
    @pditommaso
    all the config or some specific settings ?
    James Fellows Yates
    @jfy133:matrix.org
    [m]
    entire
    (screenshot coming)
    Paolo Di Tommaso
    @pditommaso
    weird, is this tower.nf? do you is there a way I can try to replicate the problem?
    James Fellows Yates
    @jfy133:matrix.org
    [m]
    yes, tower.nf
    That's actually why I asking if there are any tips to debug, because I can't find a way to replicate it
    No-one else seems to be having the problem (as far as I found so far). But it also happens to me when running locally from my laptop as well, and sent to my prviate workspace (not just the workspace/organisation)
    Paolo Di Tommaso
    @pditommaso
    anything reported in the .nextflow.log file?
    James Fellows Yates
    @jfy133:matrix.org
    [m]
    No errors or warns that I could see at least
    2 replies
    Roger Olivella
    @rolivella
    Hi all, I'm new to Nextflow Tower CLI. I'm testing it and it's very cool! I'm wondering how to get detailed info af a run. I've tried: tw runs view --id=1R5EyNgnXGTxXGbut I get ERROR: Missing the required parameter 'launchId' when calling describeLaunch. What I'm doing wrong? Thanks!
    Paolo Di Tommaso
    @pditommaso
    weird, it may be a bug. can you please report it in the Github project?
    James Fellows Yates
    @jfy133:matrix.org
    [m]

    @pditommaso: I've identified that the issue with the n/a is it happens when I activate enabled = true in the tower scope of (any) nextflow config file. It doesn't happen when I run with -with-tower.

    i.e.

    tower {
      enabled = true
      accessToken = '<YOUR TOKEN>'
      workspaceId = '<YOUR WORKSPACE ID>'
    }
    • nextflow run nf-core/eager -profile test

    will result in n/a

    but

    tower {
      accessToken = '<YOUR TOKEN>'
      workspaceId = '<YOUR WORKSPACE ID>'
    }
    • nextflow run nf-core/eager -profile test -with-tower

    works OK and I see the resolved configuration

    Should I make a bug report on the Nextflow repo instead of tower?

    James Fellows Yates
    @jfy133:matrix.org
    [m]
    (by any config file, I mean it doesn't matter if it's ~/.nextflow/config or a custom one specified with -c
    James Fellows Yates
    @jfy133:matrix.org
    [m]
    I made a report on nf-tower so I didn't forget: seqeralabs/nf-tower#338
    Roger Olivella
    @rolivella
    Done @pditommaso
    Paolo Di Tommaso
    @pditommaso
    thanks, think a patch has been already delivered
    James Fellows Yates
    @jfy133:matrix.org
    [m]
    I saw, thanks for the speed response 🙂
    daudn
    @daudn

    I have a tool that runs workflows for me, and due to a bug it ran multiple workflows all at once.

    The workflows remain in “blue” on Nextflow tower in a sort of crashed state. Unable to delete and as one of the processes is stuck as pending.

    What’s the best way around this?
    Paolo Di Tommaso
    @pditommaso
    please report an issue including the ID workflow in a faulty status
    moira-dillon
    @moira-dillon
    Hi all! I've been looking at both Cromwell and Nextflow. I know that Cromwell can automatically separate failed workflows to separate directories than successful ones - does Nextflow have that capability also?
    Phil Ewels
    @ewels
    How do you mean exactly? Nextflow only knows about one workflow run at a time.
    Tower knows about lots and reports whether they failed or not in the web interface.
    Michael L Heuer
    @heuermh
    Friends don't let friends use Cromwell :wink:
    maxulysse
    @maxulysse:matrix.org
    [m]
    😂
    Andrew Hill
    @hillan141
    Hello - when using Tower on AWS Batch, I find tasks that need to publish a file larger than ~100GB to a S3 publishDir lead to many-hour wait with repeated nextflow log message 'waiting files transfer to complete (1 files)', and ultimate failure to publish. For files smaller than ~80GB or so I have not encountered this. Appears to be the same as the reported issue on slack here: https://nfcore.slack.com/archives/CE6SDBX2A/p1629672433372600. The suggested solutions I have seen there and in related links are (1) change multipart upload size (2) use Fusion mounts in Tower compute environment. Any suggestions? Thank you.
    Paolo Di Tommaso
    @pditommaso
    nextflow has some issues transferring very large data files, we are working to improve it