No data migration. Ploomber Cloud runs on AWS, and can integrate with your existing infrastructure such as S3, GCS, RedShift and Athena.

Ploomber

<@U025MBQ6W7P> developed the Kubeflow integration, let's give it some time since he's probably asleep right now :joy:

absolutely! :slightly_smiling_face:

Thanks for the tools, having lots of fun with it.

It seems like you’re missing a client file for S3/Google cloud? This integration is counting on some cloud storage because of Kubeflow’s design

Check <https://soopervisor.readthedocs.io/en/latest/tutorials/kubeflow.html#instructions> you’ll see how to add the right dependency and configuring the bucket/keys, you should also include your clients file in the pipeline.yaml

Thanks, I run kubeflow pipelines locally, no need for aws creds. Will try adding the client file.

You can skip checking the file client. Run “soopervisor export —help” and you'll see a list of options I think it's “-s”

Also, try it first with pipeline.yaml, I don’t remember if we’ve added the training/serving yaml functionality (I believe I didn’t)

is the scheduled task running `ploomber build`? if a notebook fails, it should exit with code 1, i just tested it with some notebook. do you have a way to reproduce the error?

ok let me ask my colleague to double check. thanks

<@U032HTM8DBN> it might be easier to have them post the error/join the slack :slightly_smiling_face:

hey <@U032HTM8DBN>, any luck? I can dig into this if you are getting the wrong exit error

We switched over to looking at something else atvthe moment but don't worry about it — I assume that the exit error is as you say and we just need to capture it. Thanks.

is the problem that the loading csv file wont recognize the datetime columns? if so, you can use the parquet format

And how should that look like in my `pipeline.yaml` file.
The following is my current one:
```tasks:
  - source: load_data.py
    product: 
      nb: output/load_data.ipynb
      data: output/data.csv
      data_n: output/data_n.csv
      
  - source: find_features.py
    product: 
      nb: output/find_features.ipynb```

Just replace the “.csv” for “.parquet” and store the data frames with “df.to_parquet”

FrozenJSON is the internal representation that allows accessing dictionaries with the dot notation. I borrowed it from the book back when ploomber didn't support notebooks (it was all functions back then). If I understand correctly Ploomber is injecting a string instead of a list? This sounds like a bug. Can you provide a minimal example so I can reproduce it? 

My guess is that the problem lies in the frozenJSON to list  conversion 

i’ll see if i can put together a more minimal example than the one im using now but that should at least put you on the path. for now i can just externalize this thing outside of env.yaml. thanks

ok i have a minimal example. not in a place where it's convenient to create a ticket right now but i can give you one case that parses correctly and one that returns the internal representation 

so if i have foo: { bar: “baz” } , it shows up as expected in the notebook

if i have foo: [ { bar: “baz” } ], that shows up with the frozen json

Great! Thanks a lot for sending this! I have an idea of where the error is. I'll experiment a bit but should be fairly easy to fix. 

it seems to be that if the value is a list of complex types, it doesnt parse. a list of strings works, and a child that's a complex type works, but a list of complex types fails

create an `env.yaml` file and then reference the parameter in the tasks:

```#env.yaml
some_param: some_value

# pipeline.yaml
- source: some-task.py
  params:
    some_param: '{{some_param}}'

- source: another-task.py
  params:
    some_param: '{{some_param}}'```
does this work?

both tasks will get `some_param='some-value'`

when i use ploomber scaffold in cli, it creates a environment.yml file. I should use this?

so for example let's say I want to use the same csv file in several tasks but the csv file isn't create by anything upstream

environment.yml is to declare your conda/pip dependencies, so create an env.yaml  <https://docs.ploomber.io/en/latest/user-guide/parametrized.html|check this out>

yeah, a path to a csv is a common use case for a parameter. the recipe I shared should work. one tip, you can define the path like this:

```# env.yaml
path_to_csv: '{{here}}/path/to/data.csv'```
`{{here}}`  will be replaced for the directory where the env.yaml is located, this is more robust since you'll get an absolute path that you can load anywhere regardless of the current working directory

ah nice, that is a good tip. Just to be clear once the param is defined in the env.yaml file it needs to be re specified in the pipeline.yaml to be used in a task?

kinda preferable to go from env.yaml to task to skip retyping

yeah, you have to re-specified it. we've gotten this feedback before. any thoughts? how would you like the API to look like to be more succinct

env.yaml seems redundant, why not just specify params in pipeline.yaml?

yes {{here}} works in pipeline.yaml. sure you can define them without an env.yaml

```# pipeline.yaml
- source: some-task.py
  params:
    some_param: '{{here}}/some/path.csv'

- source: another-task.py
  params:
    some_param: '{{here}}/some/path.csv'```


yea i guess this is preferable to the env.yaml file solution

its just too easy to retype it incorrectly

once source of truth is a good design philosophy I guess

the doc u linked to is also very helpful !

Hey, thanks for your feedback! You're right, we're missing the choose-best task in the example, I'll add this to the backlog and find some time to improve it next week. PRs welcome if you wanna give it a try :slightly_smiling_face:

What version are you on? It's related to the click dependency

somehow this only happens on one of my computers, the other one is fine.

btw, Python 3.8.12 on the computer that is having this issue

please do `pip install click --upgrade` and let us know if it works

ploomber 0.18 requires click &gt;=8 to work, but this may conflict with other dependencies so in the next version (coming up in the next few days) we expanded compatibility so it works with click 7 and 8

great question, we did a survey here: <https://ploomber.io/blog/survey/>

but of course we're biased to if anyone has used kale and ploomber, feel free to weigh in :slightly_smiling_face:

This survey is inaccurate for all the tools listed in it. It does more harm to Ploomber than good since the grades are so off, people will discount the grades you gave to your tool too.

appreciate the feedback! we did this &gt;1 year ago and many things have changed, but we haven't taken the time to re-do the survey

interesting question. the benefit of reading/writing is that you "cache" results so you don't need to re-compute them, but here your bottleneck is reading/writing. parquet is a great format for fast read/write operations but I'm unsure how fast it is for text data. feather is another alternative

<https://parquet.apache.org/>
<https://arrow.apache.org/docs/python/feather.html>

please let us know if this helps!

forgot to mention: parquet is a columnar format so one way to improve read operations is to only load the columns that you're gonna use e.g. if you have 100 columns but some script only needs the first 10, you can load them. both parquet and feather work great with Python and R

this is helpful, thanks! I’ll take a look

sure, please share your experience! would love to learn how parquet/feather do with text, I've only used them with numeric data

You can also use in memory only, but then the tradeoff is that you’d have to rerun it once the process dies. Agree on parquet, not sure how performant it’ll be on plain text. It also depends on how much <https://stackoverflow.com/questions/27384093/fastest-way-to-write-huge-data-in-file|write calls> you’re performing.

yeah, so say you ran the pipeline and got all the data. next month, if you wish to re-download, you can run `ploomber build --force` and this will force execution of all tasks. alternatively, you may do `ploomber task {task-name} --force`  and force execution of that task only, next time you run `ploomber build` , ploomber will see that `{task-name}` ran recently and execute all downstream tasks. does this solve the issue?

beautiful, thats perfect! yeah, I was looking for a task specific force - this helps. Thanks!

You can use grid for that and save some typing :) 

<https://docs.ploomber.io/en/latest/cookbook/grid.html|https://docs.ploomber.io/en/latest/cookbook/grid.html>

What's the use case? Usually when you consume a python file, and execute it, it's usually the same. We have an open issue on multiple formats for user products <https://github.com/ploomber/ploomber/issues/673|https://github.com/ploomber/ploomber/issues/673>

the idea of having a notebook output is that the ipynb file can contain charts and tables in a standalone file. if it's a script, that isn't possible. what's the use case?

We wnted ti use an .ipynb file as source, generate a .py file as product, and finally import the product in another task

Is there a reason not to consume the product .ipynb as an input?
Also what you could do is running the source as a .py and then consume it (since it’s a .py there’s no context like in notebooks so you can have a generic input).

2- Yes you are right of course, but in some cases is just easier to work on the source file as a .ipynb format

Ah ok. Sounds like pairing will work for you. Pairing allows you to have two sources (py and ipynb) and keep them synced <https://docs.ploomber.io/en/latest/user-guide/editors.html|https://docs.ploomber.io/en/latest/user-guide/editors.html>

Thanks for bringing this up! So BigQuery can be used just like a db in this example if you use an ODBC/JDBC connection. You can use it securely in a similar manner to the blog (keyring). Any thoughts?

Hey <@U025MBQ6W7P>, thanks for the response!

I'm currently using sqlalchemy and the following code to connect to BigQuery:
```engine = create_engine(
    "bigquery://",
credentials_path="bigquery_credentials.json",
)```
The json file has 10 key:value pairs. I'm not sure how to proceed from here in storing the credentials securely and whether keyring is appropriate for this use case? At first glance, keyring looks to be storing username and password pairs?

Got it, so you can use the get_credential() <https://github.com/jaraco/keyring|function>, it allows you to store a full credential object for that matter that way you can store the full json and consume it. You can also use other ways to connect to it, like with drivers, or a private key.

thanks for the feedback. so right now we put all the metadata, but this makes sense. I opened an issue: <https://github.com/ploomber/ploomber/issues/735>

Screenshot at 2022-05-03 10-42-56.png

This is included in the file as well:

```# ---
# jupyter:
#   jupytext:
#     cell_metadata_filter: all
#     notebook_metadata_filter: ploomber
#     text_representation:
#       extension: .py
#       format_name: percent
#       format_version: '1.3'
#       jupytext_version: 1.13.6
#   kernelspec:
#     display_name: Python 3 (ipykernel)
#     language: python
#     name: python3
#   ploomber:
#     injected_manually: true
# ---```


It works in jupyter notebook but fails in jupyter lab

Do you have the ploomber plugin installed and the machine restarted? Was it running on different notebooks? We saw in the past users forgot to reset the instance after installation and that caused the plugin some issues.

what Ido mentioned is pretty common. to add more detail: ploomber must be installed in the process that runs jupyter (which isn't necessarily the same process that runs the kernel). although there seems so be some incompatibility here. do you know which version of ploomber and jupyter you're running?

ah, interesting feature! we don't support it yet but from what I understand is like you have a task that generates A.csv but now you renamed it to B.csv, so you want to delete A.csv?

Yup! Something like that! I’ve been doing that manually (and my pipeline right now is small enough that I can delete everything in my output directory and regenerate no problem) but it’d be neat to have a tool to do it automatically!

<@U03099A9AQ0> Would you like to open a git issue? <https://github.com/ploomber/ploomber/issues/new|https://github.com/ploomber/ploomber/issues/new>

Hey Robson! No need to apologize! Your changes haven't been merged so it's ok, no one has been affected. You can close the PR if you no longer wants us to review it. You can experiment as much as you want and open a PR when you want us to review it! This community is all about learning and helping each other!

Thanks for the understanding and the fast response. What I'm most worried about now is that /ploomber/ploomber has a banner that says "RobsonGlasscock:mybranch had recent pushes x minutes ago." Is there a way this can be undone?

so this message is saying you pushed to your fork, but that doesn't affect the main repo (although I agree the message is confusing since it's appearing in /ploomber/ploomber)

you can undone it, by deleting the commits but there is no need to, the changes are still in your fork

Okay. Thank you for the explanation. It really threw me off seeing it in /ploomber/ploomber. Thanks also for the patience. I know y'all are busy.

Feel free to ask more, I think it’ll be easiest for you to open a new branch, cherry pick your changes and resubmit :slightly_smiling_face:

Appreciate that, and the new PR is submitted.

Haha, I had a similar thing happen in one of my earlier contributions. <@U0164GZJ5EK> and <@U025MBQ6W7P> and community organizers are very helpful and have a good process for ploomber. Welcome to ploomber and don't be afraid to ask any questions! :slightly_smiling_face:

A few thoughts here, are you using jupyter locally? this error seems unrelated to the package itself.
Did you try running one of the samples? It’ll be easier to understand where’s the issue that way, since if it’ll run fine it’s not with core ploomber but probably something else is causing it

I am using Ploomber on aws sagemaker instance

Earlier it was working with one of my files, it was basic file only

Finally, if you have a specific notebook that’s already running fine and you just want to convert it into a pipeline, you can use the <https://github.com/ploomber/soorgeon|soorgeon tool>, it’s pretty intuitive - all you need to do is set some H2 headings on the notebooks.

Yes, I converted using sooreon tool and spilt the notebooks into the pipelines

Can you try running the <https://github.com/ploomber/projects/tree/master/templates/ml-basic|ml-basic> example? We had a few users that were running on Sagemaker without issues

Ok so soorgeon ran and now you have a modular pipeline? Was the notebook running fine before it?

Also, let me know how the sample execution is going

Nothing is running it throws the error which I shared

Even with this? <https://github.com/ploomber/projects/tree/master/templates/ml-basic>

`ploomber examples -n templates/ml-basic`

This I have not tried, I was building with my ipynb file

Ok so please try this first it’ll give you an insight if the error was the conversion to the pipeline or the environment itself.

so this looks like a kernel issue: your ipynb file contains kernel information (<https://papermill.readthedocs.io/en/latest/troubleshooting.html|see here>), but it looks like now that kernel no longer exists or it isn't available from where you're running things. I'd recommend either modifying the kernel spec metadata manually or creating kernel with the name that appears in the error message `conda_python3`

Thanks Eduardo, how to modify the kernel specs

Hello Edardo and Ido, can you please help me here,

I was able to create the orceshestor file using sorrgeon, however it Ploomber fails to build with the modified elements,

==================================================== DAG build failed =====================================================
--------------- NotebookRunner: import-the-libraries -&gt; MetaProduct({'nb': File('output/...raries.ipynb')}) ---------------
-------------------------------- /home/ec2-user/SageMaker/tasks/import-the-libraries.ipynb --------------------------------
---------------------------------------------------------------------------
Exception encountered at "In [1]":
---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
&lt;ipython-input-1-da6c937aa8cd&gt; in &lt;module&gt;
      1 import tensorflow as tf
----&gt; 2 from tensorflow.keras.layers import log, e

ImportError: cannot import name 'log' from 'tensorflow.keras.layers' (/home/ec2-user/anaconda3/envs/JupyterSystemEnv/lib/python3.7/site-packages/keras/api/_v2/keras/layers/__init__.py)

ploomber.exceptions.TaskBuildError: Error when executing task 'import-the-libraries'. Partially executed notebook available at /home/ec2-user/SageMaker/output/import-the-libraries.ipynb
ploomber.exceptions.TaskBuildError: Error building task "import-the-libraries"
==================================================== Summary (1 task) =====================================================
NotebookRunner: import-the-libraries -&gt; MetaProduct({'nb': File('output/...raries.ipynb')})
==================================================== DAG build failed =====================================================

Did you check if the code runs without ploomber? It looks like a missing import in a specific task “import-the-libraries”

hey. yes, if you have any suggestions (like the pre commits), feel free to open an issue to discuss

thanks for your patience, i'll review your PR in the next few days!

Yes, if you are using a pipeline.yaml, pass a dictionary in product, then the `product` argument inside the function will be a dictionary too

if using the Python API, you can also pass a dict (<https://github.com/ploomber/projects/blob/c385f55d611a4053bc423c4e608a46c00d5a942a/templates/python-api/src/ploomber_basic/pipeline.py#L37|see here> - note that this is a Notebook task but it works the same with PythonCallable)

yep, yep, but how about using the Python API? as per the docs seems only one product is possible?

yep: <https://docs.ploomber.io/en/latest/api/spec.html>

how did you search? (interested in learning if you tried looking it up and didn't find this document)

aha, so <https://docs.ploomber.io/en/latest/api/spec.html#schema> in particular.

Yes, I searched and read results for
• `DAGSpec` 
• `meta` 
and additionally reviewed
• <https://docs.ploomber.io/en/latest/user-guide/parametrized.html> 
• <https://docs.ploomber.io/en/latest/user-guide/configuration.html> 

ah, so looks like our search engine isn't indexing the content of the snippets but only the text. thanks for the feedback!

there's an open issue about it :slightly_smiling_face: <@U032K7U1E6L> is working on it, hoping to merge this soon

Yes, hoping to have something for this soon

great question! there isn't one particular feature to link the pipelines but you can achieve this with <https://docs.ploomber.io/en/latest/user-guide/parametrized.html|parametrized pipelines>. For example, you can define an `output_i_want_to_link` key in your `env.yaml`  and map it to the output location (say `clean-data.csv` ) then reference it in `pipeline.preprocess.yaml`  (as a `product` in the final task) and `pipeline.model.yaml`  (as a `param` in the first task) with `{{output_i_want_to_link}}` , this way, you'll avoid hardcoding the path. to mark the training pipeline as outdated when your data changes you can use <https://docs.ploomber.io/en/latest/api/spec.html#tasks-params-resources|resources>_ feature. however, this will cause ploomber to compute the hash on the file, which isn't scalable if the file is too big. what's the file size?

also, out of curiosity. why not having a single `pipeline.yaml`?

one file is ~77.5mb and the other is ~25mb….so not horrible but definitely not teeny tiny!

ah, I think using `resources_`  will work. It'll probably take a few seconds to hash the file but it will solve your problem. let me know how it goes. we have a long-standing issue about  adding a feature to facilitate composing pipelines and maybe it's the right time to tackle it :slightly_smiling_face:

re: multiple pipelines, we’re getting to the point where we have ~15 tasks for the preprocessing steps and ~10 tasks in the modeling steps. We can keep them in one pipeline, but we’ve been conceptualizing them as two major steps. Plus, most of our collaborators and other scientists only realllly care about (for better or for worse) the initial output after we do all our data munging and cleaning plus the model results :rolling_on_the_floor_laughing:

ah, this makes sense. I didn't think about it this way before. I'm guessing it makes it simpler for each collaborator since they don't need to deal with understanding other parts of the pipeline.

I think my suggestion will work but feel free to send any other questions!

Hey! Just working on this now, I’ve defined the parameter in my `env.yml`, but I’m getting a warning that its not defined…how do I check which env ploomber is pointing to?

```Error: Error replacing placeholders:
  * {{save_path}}: Ensure the placeholder is defined in the env

Loaded env: EnvDict({'cwd': '/Users/jessi...ense_pipeline', 'git': 'main', 'git_hash': 'c5441e3-dirty', 'here': '/Users/jessi...ense_pipeline', ...})```

and my env.yml def has
```save_path: preprocessing/output/processed_data/preprocessed_semcor_tags.csv```

and my pipeline.yml has
```- source: preprocessing/scripts/preprocess_semcor_tags.py
  product:
    nb: preprocessing/output/notebooks/preprocess_semcor_tags.ipynb
    data: {{save_path}}```

I’m worried its finding a different env or something? I’d just like to check which env its parsing :thinking_face:

Ah, it’s because the name should be env.yaml, rename it and it should work!

Ha! I knew it’d be something simple, thanks!

Sure. I’d recommend adding {{root}}/preprocessing/… prefixing it will convert it to an absolute path and it will ensure you can load the file from anywhere

following your example are you looking to upload `my.log`? if so, where do you want to upload it?

S3 bucket. But the log are generated after I runned the pipeline.

ah. this is an interesting use case. you'd have to add code to manually upload the log file to S3, but since we already support S3 for uploading artifacts. it makes sense to support uploading the log as well. can you create an issue?

If you show me how I will gladly create an issue :) 

I can always create another pipeline to upload various log file created while running ETL pipeline and run that one when needed meanwhile. Uploading log to bucket would be for trackability and keeping record of performance. We can't never have too much data :the_horns::sweat_smile:

sure, create the issue here: <https://github.com/ploomber/ploomber/issues/new>

This seems like an issue with the conda env environment in general.  I haven't use ipython extensions heavily but my guess is that your ipython copy in the `analysis`  env is finding your `profile_default` and then trying to load the extensions. I can think of a couple options:

1. disable the extensions globally
2. disable them locally (not sure how to do this but perhaps you can create a `jupyter/.ipython/profile_default` that is local to the analysis conda env
3. install the sql extensions (and any others) in the conda env
thoughts?

Looking at ploomber’s <https://github.com/ploomber/ploomber/blob/d2ef3276fad0f871bf2677b7fc592f35b0bd980d/src/ploomber/cli/interact.py#L33|interact.py>, I see it calls start_ipython and links to this <https://github.com/ipython/ipython/issues/8918#issuecomment-149905901|ipython issue>

One option could be to set profile to none in start_ipython config - though I couldn’t find any example of that.

I can’t disable/change default profile as that is used for the notebook interface.

I understand this is probably a low priority issue to solve, though here are my thoughts:

Is loading default config a good thing? I would say no. As default config is meant for default environment - whereas in production almost always ploomber will not be running in default environment. In such a scenario I can see 3 options:
• Ploomber starts ipython in a way so that no profile gets loaded (either settting argv or config in start_ipython)
• Ploomber ships with its own profile and uses that.
• There is some command line or config param which lets a user specify profile to use with ploomber.


Ah interesting. Thanks for sending this, I'll dig into this a bit more. Does this mean Ploomber works well except for “Ploomber Interact”? Or is the whole installation broken?

I tried ploomber build with one task with ipynb source. It actually ran and completed the build successfully even though it did throw errors about missing ipython modules.

thanks for the update! was it the same error as before (fail to load sql  module)? - I'll open an issue on github so we keep track of this. i agree with your earlier arguments about isolating ploomber from the system wide config

Yes. It printed warnings on two modules. One being sql and other being a private package beatrix-jupyterlab on gcp vertex ai workbench instances:

```❯ pip show beatrix-jupyterlab
Name: beatrix-jupyterlab
Version: 3.1.7
Summary: Beatrix Notebooks extension package
Home-page: <http://go/notebooks-frontend>
Author: Vertex AI Notebooks FE Team
Author-email: <mailto:cloud-ai-frontend-notebooks@google.com|cloud-ai-frontend-notebooks@google.com>
License: Apache License 2.0
Location: /opt/conda/lib/python3.7/site-packages
Requires: google-auth, google-cloud-bigquery, google-cloud-bigquery-storage, google-cloud-storage, jinja2, jupyter-server, jupyterlab, matplotlib, pandas, pexpect, psutil, pyarrow, pyjwt, requests, tqdm
Required-by:```

great, thanks for the feedback! I'll look into this to see how we can fix it. feel free to send any other questions!

hi. here's an example to use cloud storage <https://docs.ploomber.io/en/latest/api/_modules/clients/ploomber.clients.GCloudStorageClient.html#ploomber.clients.GCloudStorageClient>

for big query, are you looking to run a query and dump the results into a local file or do you want to write SQL scripts that create new tables?

we're still working on improving these examples so please share your feedback!

my use case - read a kafka topic, store output to GCS, read GCS and update bigquery table after flattening it, transformation using SQL and create a materialized view.

I saw the documentation. I was hoping you have a github repo of working example of using Bigquery and GCS.

ah. we don't have a full example. thanks for the suggestion! I think it makes sense to add one. Also for AWS

but feel free to ask any questions here as you make progress with the gcloud integration

<@U0164GZJ5EK> - here is the error I’m getting. ModuleNotFoundError: An error occured when trying to import dotted path ‘clients.gcs’: No module named ‘clients’

do you have a `clients.py` file in the same folder as your `pipeline.yaml` ?looks like it cannot find it

thx. now it is ModuleNotFoundError: No module named ‘google’

probably missing the g cloud storage package? <https://pypi.org/project/google-cloud-storage/>

thx got it working. is there a way to parameterize clients.py? hate to use path hardcoded in .py file.

image.png

in your case it'd be something like:
```File:
  dotted_path: clients.gcs
  name: value```

hi <@U0370FZQZ7E>! I was looking for ideas for our next blog post and I decided to write something on using ploomber with google cloud :slightly_smiling_face: since you're working on setting this up, I thought of sharing the <https://github.com/ploomber/projects/tree/master/templates/google-cloud|code example I'm working on> - right now, it connects to bigquery, but I'll add the cloud storage stuff too. still work in progress but feel free to share your feedback

Do you plan to use sqlalchemy client or DBAPIClient or native BigQuery Client?

DBAPIPClient is a wrapper for the native client. using the sqlalchemy client has a few extra features, but both work fine

we don't have scheduling at the moment. what teams do is they call `ploomber build` from a cron jon (or similar) or <https://soopervisor.readthedocs.io/en/latest/|export their pipelines> to an orchestrator that has scheduling (argo, airflow)

thanks. are you using AWS of Google as cloud provider? will clients have a choice? the reason is we have huge data in Google Cloud so it makes sense to execute pipeline in it to limit egress charges and IO.

so right now our cloud services are built on AWS. we are planning to support Google Cloud as well but that will mostly depend on client's demand. So for current teams on Google Cloud or Azure, we are helping them build their infra on their existing cloud provider using our open source tools

This is some of the plugins we used, there was a story about removing this, I just pushed a fix, please report if that happens again :slightly_smiling_face:

Feel free to update how the cloud works for you, happy to help!

Did you add H2 headers for both tasks? Are those imports being used?

import-the-libraries - what’s the purpose of the task just to do imports?

So you mean to say that import libraries in one task ?

I meant it seemed the task that was failing was called import-the-libraries and it seemed like a missing dependency. Were you able to fix it?