These are chat archives for nextflow-io/nextflow

26th
Feb 2019
Jorrit Boekel
@glormph
Feb 26 09:42
Issue filed. Am a bit uncertain if this is actually a bug: nextflow-io/nextflow#1048
Or my own problematic pipeline programming :)
Paolo Di Tommaso
@pditommaso
Feb 26 10:38
it looks doesn't look like a bug
Jorrit Boekel
@glormph
Feb 26 10:41
I thought this was the test case?
Isolated the problem in the pipeline to these lines, buffering a large amount of input?
Alan B. Christie
@alanbchristie
Feb 26 10:56
I'm trying to use the serviceAccount feature of the k8s config in an OpenShift environment. Pods in the namespace I'm running in cannot be launched or watched by the default SA. When I set the serviceAccount in the config the default account is still being used. I see this from nextflow: pods is forbidden: User "system:serviceaccount:squonk:default" cannot create pods in the namespace "squonk". Nextflow claims to have found the local configuration file which I know contains the line serviceAccount = 'squonk' so I would expect the user to be system:serviceaccount:squonk:squonk. Is there a way I can debug nextflow to understand why it's using the default account rather than the one defined in the config? At least to eliminate any error I have made.
Alan B. Christie
@alanbchristie
Feb 26 11:08
...using nextflow 18.10.1
Paolo Di Tommaso
@pditommaso
Feb 26 11:19
put in the config
    k8s {
      debug.yaml = true
    }
then check prod spec created
and eventually open a bug
Alan B. Christie
@alanbchristie
Feb 26 11:20
Will do
Alan B. Christie
@alanbchristie
Feb 26 11:46
OK - initial fault was the SA of the nextflow main Pod. I'm past the error and things are working but the pod spec for the spawned pods still uses the default service account. Although this is not causing me problems atm I'll raise an issue to track it. Thanks.
k-hench
@k-hench
Feb 26 15:02

Hi everyone,

I have a quite general question for which I could not find the answer in the nextflow docs:

Is there any possibility to run several nextflow scripts in the same folder in parallel?

Currently I run into issues when trying, I guess because both processes try to access the .nextflow folder simultaneously.
I don't find a way to adjust this (eg. like when renaming the report files).

cheers,
Kosmas

Luca Cozzuto
@lucacozzuto
Feb 26 15:03
Hi @k-hench, I suggest you to run the nextflow scripts in different folders
so that you won't have any collision
you can always point to a single folder for input data
Maxime Garcia
@MaxUlysse
Feb 26 15:05
I would suggest the same, otherwise, you can try to have the report.file to have a name based on the current run if you really need to run in the same folder
k-hench
@k-hench
Feb 26 15:15
Thanks for the advice :)
I guess I need to keep things separate then...
najitaleb
@najitaleb
Feb 26 19:08
Hi. I just wanted to know if there was any update or solution to the problem mentioned at the end of this discussion. (udocker.py: command not found) Thanks. https://groups.google.com/forum/#!topic/nextflow/ZHag7dQhCN0
Joseph Guhlin
@jguhlin
Feb 26 21:35
Hello. I'm working with conda, but nextflow's bash doesn't process login scripts so conda isn't found unless I start bash before running nextflow (I use fish, and conda works no problem for me). Is there any way to force nextflow to bash -l (or some other flag)?
Anthony Underwood
@aunderwo
Feb 26 23:39

Hi all - please can anybody advise on how to import java functionality into a .nf file

I have exported the aws sdk paths as follows

export NXF_CLASSPATH=/home/ubuntu/aws-java-sdk/third-party/lib:/home/ubuntu/aws-java-sdk/lib

but I get the error

_nf_script_c71fadad: 6: unable to resolve class com.amazonaws.services.dynamodbv2.model.GetItemRequest
 @ line 6, column 1.
   import com.amazonaws.services.dynamodbv2.model.GetItemRequest;
   ^