Fix missing curl binary and update CI

[AndreMiras]

Commit summary

Fix parquet file download error because of a missing curl binary. This was happening on the eth_getBalance test for instance. Note that flood swalled the exception making the error harder to spot.

Update the CI to also test against eth_getBalance which triggers a parquet file download.
Also removes the redundant node2, this was a workaround as flood used to crash with only one node.

Also unpin the patch version of Python and address a small DRY on the Vegeta version.

Motivation

As in version 0.3.1, eth_getBalance fails in a docker environment:

docker run --rm paradigmxyz/flood:latest eth_getBalance node1=https://eth.llamarpc.com --duration 3 --rate 1

Output:

┌───────────────────────────┐
│ Load test: eth_getBalance │
└───────────────────────────┘
- sample rates: [1]
- sample duration: 3
- extra args: None
- output directory: /tmp/tmpsrenaj0e

Gathering node data...
──────────────────────
[WARNING] ctc config file does not exist; use `ctc setup` on command line to generate a config file

       node  │                       url  │               metadata  
    ─────────┼────────────────────────────┼─────────────────────────
      node1  │  https://eth.llamarpc.com  │  llamanodes_web3_proxy  
             │                            │               v1.42.11  


Running load tests...
─────────────────────
[2023-08-30 20:37:24] Starting
[2023-08-30 20:37:24] Running load test for node1
/home/flood/.local/lib/python3.11/site-packages/flood/generators/raw_data_sources/raw_download_utils.py:68: UserWarning: FLOOD_SAMPLES_DIR not set in env, defaulting to CWD
  warnings.warn('FLOOD_SAMPLES_DIR not set in env, defaulting to CWD')
downloading ethereum_contracts_samples__L__v1_0_0.parquet

downloading https://datasets.paradigm.xyz/evm_samples/ethereum_samples__v1_0_0/ethereum_contracts_samples__L__v1_0_0.parquet
2

Solution

Fix parquet file download error because of a missing curl binary. This was happening on the eth_getBalance test for instance. Note that flood swallowed the exception making the error harder to spot.

PR Checklist

• Added Tests
• Added Documentation
• Breaking changes

[sslivkoff]

👍

so just to be clear, there is no actual error with the parquet file right? this new image just embeds the parquet file inside the image for convenience?

[AndreMiras]

Not exactly, it's happening while trying to download the parquet file using the pdp utility package here:
https://github.com/paradigmxyz/flood/blob/0.3.1/flood/generators/raw_data_sources/raw_download_utils.py#L58
However pdp relies on curl here https://github.com/paradigmxyz/paradigm-data-portal/blob/0.2.2/pdp/data_utils/file_utils.py#L48-L57
So without that change, running the eth_getBalance test in a container fails with a swallowed FileNotFoundError.
The non-swallowed exception would be something like:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/flood/.local/lib/python3.11/site-packages/pdp/data_utils/file_utils.py", line 57, in download_file
    subprocess.call(['curl', url, '--output', output_path])
  File "/usr/local/lib/python3.11/subprocess.py", line 389, in call
    with Popen(*popenargs, **kwargs) as p:
         ^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/subprocess.py", line 1024, in __init__
    self._execute_child(args, executable, preexec_fn, close_fds,
  File "/usr/local/lib/python3.11/subprocess.py", line 1917, in _execute_child
    raise child_exception_type(errno_num, err_msg, err_filename)
FileNotFoundError: [Errno 2] No such file or directory: 'curl'

Edit:
So to clarify, this image doesn't embed the parquet file, that file is still being downloaded by flood when not available.
This is that exact download process that was failing because the curl binary was missing from the container.
This parquet download was introduced in b7c98b2, so we're basically just missing curl

[sslivkoff]

oh. gotcha. lgtm

[AndreMiras]

This is the bug fix, the rest is just cosmetic