Fix high CPU usage during ramalama pull #685

mkesper · 2025-01-31T14:53:49Z

Feel free to adjust block size. But 1k blocks are ridiculously low for downloading GiB of LLM data.

Related: #684

Summary by Sourcery

New Features:

Added a --quiet flag to suppress the progress bar during downloads.

Showing update progress in 1 kiB blocks adds unnecessary overhead. 100 kiB or even 1 MiB should be sufficient. Related: containers#684

Related: containers#684

sourcery-ai · 2025-01-31T14:53:55Z

Reviewer's Guide by Sourcery

This pull request introduces a new --quiet flag to suppress progress bars during downloads, and increases the download block size to 100KB to improve download speeds and reduce CPU usage.

Sequence diagram for the modified download process

sequenceDiagram
    participant Client
    participant DownloadManager
    participant HTTPClient

    Client->>DownloadManager: download_file(url, show_progress)
    DownloadManager->>HTTPClient: init(url, show_progress)
    activate HTTPClient
    loop Until EOF
        HTTPClient->>HTTPClient: read(100KB blocks)
        Note right of HTTPClient: Increased from 1KB to 100KB blocks
        alt show_progress is true
            HTTPClient->>Client: Update progress bar
        end
    end
    HTTPClient-->>DownloadManager: Download complete
    deactivate HTTPClient
    DownloadManager-->>Client: File downloaded

Class diagram showing modified download components

classDiagram
    class HTTPClient {
        -response
        -now_downloaded: int
        -start_time: float
        +init(url, headers, output_file, show_progress)
        +perform_download(file, progress)
    }

    class DownloadManager {
        +download_file(url, dest_path, headers, show_progress)
    }

    class PullCommand {
        +pull(args)
    }

    PullCommand ..> DownloadManager : uses
    DownloadManager ..> HTTPClient : uses

    note for HTTPClient "Block size increased to 100KB"

File-Level Changes

Change	Details	Files
Added a `--quiet` flag to suppress progress bars during downloads.	Added a `--quiet` argument to the `pull_parser` in `ramalama/cli.py`. Modified `ramalama/url.py` to pass the `quiet` argument to the `download_file` function. Modified `ramalama/oci.py` to pass the `quiet` argument to the container engine.	`ramalama/cli.py` `ramalama/url.py` `ramalama/oci.py`
Increased the download block size to 100KB.	Changed the read size in `perform_download` from 1024 bytes to 100 * 1024 bytes.	`ramalama/http_client.py`
Modified `download_file` to accept a `show_progress` argument.	Modified `download_file` in `ramalama/common.py` to accept a `show_progress` argument. Modified `init_pull` and `pull_blob` in `ramalama/ollama.py` to pass the `show_progress` argument to `download_file`.	`ramalama/common.py` `ramalama/ollama.py`

Tips and commands

Interacting with Sourcery

Trigger a new review: Comment @sourcery-ai review on the pull request.
Continue discussions: Reply directly to Sourcery's review comments.
Generate a GitHub issue from a review comment: Ask Sourcery to create an
issue from a review comment by replying to it. You can also reply to a
review comment with @sourcery-ai issue to create an issue from it.
Generate a pull request title: Write @sourcery-ai anywhere in the pull
request title to generate a title at any time. You can also comment
@sourcery-ai title on the pull request to (re-)generate the title at any time.
Generate a pull request summary: Write @sourcery-ai summary anywhere in
the pull request body to generate a PR summary at any time exactly where you
want it. You can also comment @sourcery-ai summary on the pull request to
(re-)generate the summary at any time.
Generate reviewer's guide: Comment @sourcery-ai guide on the pull
request to (re-)generate the reviewer's guide at any time.
Resolve all Sourcery comments: Comment @sourcery-ai resolve on the
pull request to resolve all Sourcery comments. Useful if you've already
addressed all the comments and don't want to see them anymore.
Dismiss all Sourcery reviews: Comment @sourcery-ai dismiss on the pull
request to dismiss all existing Sourcery reviews. Especially useful if you
want to start fresh with a new review - don't forget to comment
@sourcery-ai review to trigger a new review!
Generate a plan of action for an issue: Comment @sourcery-ai plan on
an issue to generate a plan of action for it.

Customizing Your Experience

Access your dashboard to:

Enable or disable review features such as the Sourcery-generated pull request
summary, the reviewer's guide, and others.
Change the review language.
Add, remove or edit custom review instructions.
Adjust other review settings.

Getting Help

Contact our support team for questions or feedback.
Visit our documentation for detailed guides and information.
Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.

sourcery-ai

Hey @mkesper - I've reviewed your changes - here's some feedback:

Overall Comments:

The --quiet flag should use action="store_true" instead of action="store" with default=False to be consistent with other boolean flags in the codebase

Here's what I looked at during the review

🟡 General issues: 1 issue found
🟢 Security: all looks good
🟢 Testing: all looks good
🟢 Complexity: all looks good
🟢 Documentation: all looks good

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}

sourcery-ai · 2025-01-31T14:54:47Z

ramalama/ollama.py

@@ -45,9 +45,9 @@ def pull_blob(repos, layer_digest, accept, registry_head, models, model_name, mo
    run_cmd(["ln", "-sf", relative_target_path, model_path])


-def init_pull(repos, accept, registry_head, model_name, model_tag, models, model_path, model):
+def init_pull(repos, accept, registry_head, model_name, model_tag, models, model_path, model, show_progress):


issue (bug_risk): Function signature mismatch: pull_config_blob is called with show_progress parameter but doesn't accept it

ericcurtin · 2025-02-01T10:17:13Z

Change looks fine to me @mkesper you just need to sign your commits with something like:

git commit -a -s --amend

to satisfy the DCO bot.

ericcurtin · 2025-02-01T10:50:44Z

Lots of tests failed too, we need to get this build green somehow, at least some of the failures are flakes, not sure how many.

This will all be rebuilt anyway when you the commits are signed.

jhjaggars · 2025-02-01T16:28:17Z

ramalama/common.py

@@ -177,7 +177,7 @@ def download_file(url, dest_path, headers=None, show_progress=True):
        show_progress = False

    try:
-        http_client.init(url=url, headers=headers, output_file=dest_path, progress=show_progress)
+        http_client.init(url=url, headers=headers, output_file=dest_path, show_progress=show_progress)


This rename needs to be made to HttpClient.init as well.

ericcurtin · 2025-02-01T16:57:23Z

Anybody know what "podman pull" or "podman artifact pull" block size is? @baude @rhatdan

We could just standardise on that, it's likely a sane value.

I probably care about reliability over cpu usage. I know in practice at the TCP level of the stack a max packet size is 1500 bytes, but that's probably not so important at the http level.

ericcurtin · 2025-02-01T16:59:04Z

It could be the progress bar causing the CPU usage issue also, maybe we could change it so that it doesn't update every single block iteration.

ericcurtin · 2025-02-01T17:30:31Z

Does quiet alone lower the CPU usage significantly @mkesper ?

jhjaggars · 2025-02-03T01:00:04Z

@ericcurtin Best I can tell, the image copier used in podman updates once per second by default:
https://github.com/containers/podman/blob/main/vendor/github.com/containers/common/libimage/copier.go#L315

ericcurtin · 2025-02-03T01:55:44Z

If you guys could check if this makes a difference I'd appreciate it, not much CPU usage issues on my machine (and maybe the time calculation is as expensive that printing a progress bar update, I've no idea):

#717

mkesper added 2 commits January 31, 2025 15:31

Read data in 100 kiB blocks

3f90013

Showing update progress in 1 kiB blocks adds unnecessary overhead. 100 kiB or even 1 MiB should be sufficient. Related: containers#684

Add cli flag to suppress pull progress

94bd95e

Related: containers#684

sourcery-ai bot reviewed Jan 31, 2025

View reviewed changes

ericcurtin approved these changes Feb 1, 2025

View reviewed changes

jhjaggars reviewed Feb 1, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix high CPU usage during ramalama pull #685

Fix high CPU usage during ramalama pull #685

mkesper commented Jan 31, 2025 •

edited by sourcery-ai bot

Loading

sourcery-ai bot commented Jan 31, 2025 •

edited

Loading

Interacting with Sourcery

Customizing Your Experience

Getting Help

sourcery-ai bot left a comment

sourcery-ai bot Jan 31, 2025

ericcurtin commented Feb 1, 2025

ericcurtin commented Feb 1, 2025 •

edited

Loading

jhjaggars Feb 1, 2025

ericcurtin commented Feb 1, 2025 •

edited

Loading

ericcurtin commented Feb 1, 2025

ericcurtin commented Feb 1, 2025 •

edited

Loading

jhjaggars commented Feb 3, 2025

ericcurtin commented Feb 3, 2025 •

edited

Loading

Fix high CPU usage during ramalama pull #685

Are you sure you want to change the base?

Fix high CPU usage during ramalama pull #685

Conversation

mkesper commented Jan 31, 2025 • edited by sourcery-ai bot Loading

Summary by Sourcery

sourcery-ai bot commented Jan 31, 2025 • edited Loading

Reviewer's Guide by Sourcery

Sequence diagram for the modified download process

Class diagram showing modified download components

File-Level Changes

Interacting with Sourcery

Customizing Your Experience

Getting Help

sourcery-ai bot left a comment

Choose a reason for hiding this comment

sourcery-ai bot Jan 31, 2025

Choose a reason for hiding this comment

ericcurtin commented Feb 1, 2025

ericcurtin commented Feb 1, 2025 • edited Loading

jhjaggars Feb 1, 2025

Choose a reason for hiding this comment

ericcurtin commented Feb 1, 2025 • edited Loading

ericcurtin commented Feb 1, 2025

ericcurtin commented Feb 1, 2025 • edited Loading

jhjaggars commented Feb 3, 2025

ericcurtin commented Feb 3, 2025 • edited Loading

mkesper commented Jan 31, 2025 •

edited by sourcery-ai bot

Loading

sourcery-ai bot commented Jan 31, 2025 •

edited

Loading

ericcurtin commented Feb 1, 2025 •

edited

Loading

ericcurtin commented Feb 1, 2025 •

edited

Loading

ericcurtin commented Feb 1, 2025 •

edited

Loading

ericcurtin commented Feb 3, 2025 •

edited

Loading