Log output from serving models for easier debugging #18
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
bobcatfish
force-pushed
the
loggy_logs
branch
2 times, most recently
from
2466906
to
d817835
Compare
bobcatfish
added a commit
to bobcatfish/localllm-gcp
that referenced
this pull request
Feb 15, 2024
llama-cpp-python has a flag 'verbose' which defaults to true and when set causes it to write things to stderr. It doesn't include anyway to configure where these logs are directed, so it's stderr or nothing. Unfortunately the when we start the process running llama-cpp-python, we provide a pipe for stderr and then promptly close it. This means if llama-cpp-python tries to write to stderr, a broken pipe exception is thrown, which for example happens if there is a prefix cache hit when processing a prompt (https://github.com/abetlen/llama-cpp-python/blob/ae71ad1a147b10c2c3ba99eb086521cddcc4fad4/llama_cpp/llama.py#L645) which likely explains why ppl are seeing 500s on the second time that they try to run the same prompt. There are other situations that can make llama-cpp-python try to write to stderr as well, which may also cause 500s The real fix here is to a) not provide a broken pipe for stderr and b) for llama-cpp-python to allow us to configure logs (GoogleCloudPlatform#18). For now we can disable verbose mode in llama-cpp-python since we're not making those logs available anyway and it should stop the 500s. Fixes GoogleCloudPlatform#7
bobcatfish
force-pushed
the
loggy_logs
branch
from
d817835
to
0a9f975
Compare
bobcatfish
added a commit
to bobcatfish/localllm-gcp
that referenced
this pull request
Feb 15, 2024
llama-cpp-python has a flag 'verbose' which defaults to true and when set causes it to write things to stderr. It doesn't include anyway to configure where these logs are directed, so it's stderr or nothing. Unfortunately the when we start the process running llama-cpp-python, we provide a pipe for stderr and then promptly close it. This means if llama-cpp-python tries to write to stderr, a broken pipe exception is thrown, which for example happens if there is a prefix cache hit when processing a prompt (https://github.com/abetlen/llama-cpp-python/blob/ae71ad1a147b10c2c3ba99eb086521cddcc4fad4/llama_cpp/llama.py#L645) which likely explains why ppl are seeing 500s on the second time that they try to run the same prompt. There are other situations that can make llama-cpp-python try to write to stderr as well, which may also cause 500s The real fix here is to a) not provide a broken pipe for stderr and b) for llama-cpp-python to allow us to configure logs (GoogleCloudPlatform#18). For now we can disable verbose mode in llama-cpp-python since we're not making those logs available anyway and it should stop the 500s. Fixes GoogleCloudPlatform#7
This was referenced Feb 15, 2024
jerop
reviewed
Feb 16, 2024
llama-cpp-python uses uvicorn and it turns out there's a different way to start the running models that uses uvicorn directly, and that makes it possible to pass logging configuration to uvicorn. Unfortunately it seems that logs that come directly from llama_cpp get thrown to stderr and it's not configurable (https://github.com/abetlen/llama-cpp-python/blob/ae71ad1a147b10c2c3ba99eb086521cddcc4fad4/llama_cpp/_logger.py#L30) This change will make it so that the llm tool is installed in the workstations impage with a default logging configuration file which writes logs to /var/log/localllm.log. I don't love how different the story is if you actually run the tool directly (you have to go out of your way to get the logging) but this seems okay for now at least. The content of the logging config is from https://gist.github.com/liviaerxin/d320e33cbcddcc5df76dd92948e5be3b Fixes GoogleCloudPlatform#16
bobcatfish
force-pushed
the
loggy_logs
branch
from
0a9f975
to
9aab71f
Compare
bobcatfish
added a commit
to bobcatfish/localllm-gcp
that referenced
this pull request
Feb 16, 2024
llama-cpp-python has a flag 'verbose' which defaults to true and when set causes it to write things to stderr. It doesn't include anyway to configure where these logs are directed, so it's stderr or nothing. Unfortunately the when we start the process running llama-cpp-python, we provide a pipe for stderr and then promptly close it. This means if llama-cpp-python tries to write to stderr, a broken pipe exception is thrown, which for example happens if there is a prefix cache hit when processing a prompt (https://github.com/abetlen/llama-cpp-python/blob/ae71ad1a147b10c2c3ba99eb086521cddcc4fad4/llama_cpp/llama.py#L645) which likely explains why ppl are seeing 500s on the second time that they try to run the same prompt. There are other situations that can make llama-cpp-python try to write to stderr as well, which may also cause 500s The real fix here is to a) not provide a broken pipe for stderr and b) for llama-cpp-python to allow us to configure logs (GoogleCloudPlatform#18). For now we can disable verbose mode in llama-cpp-python since we're not making those logs available anyway and it should stop the 500s. Fixes GoogleCloudPlatform#7
bobcatfish
added a commit
to bobcatfish/localllm-gcp
that referenced
this pull request
Feb 16, 2024
Addressing feedback from @jerop in GoogleCloudPlatform#18: - Making the docstring more accurate for listing running processes - Making the check if log_config is provided more python-y
jerop
pushed a commit
that referenced
this pull request
Feb 16, 2024
llama-cpp-python has a flag 'verbose' which defaults to true and when set causes it to write things to stderr. It doesn't include anyway to configure where these logs are directed, so it's stderr or nothing. Unfortunately the when we start the process running llama-cpp-python, we provide a pipe for stderr and then promptly close it. This means if llama-cpp-python tries to write to stderr, a broken pipe exception is thrown, which for example happens if there is a prefix cache hit when processing a prompt (https://github.com/abetlen/llama-cpp-python/blob/ae71ad1a147b10c2c3ba99eb086521cddcc4fad4/llama_cpp/llama.py#L645) which likely explains why ppl are seeing 500s on the second time that they try to run the same prompt. There are other situations that can make llama-cpp-python try to write to stderr as well, which may also cause 500s The real fix here is to a) not provide a broken pipe for stderr and b) for llama-cpp-python to allow us to configure logs (#18). For now we can disable verbose mode in llama-cpp-python since we're not making those logs available anyway and it should stop the 500s. Fixes #7
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
llama-cpp-python uses uvicorn and it turns out there's a different way to start the running models that uses uvicorn directly, and that makes it possible to pass logging configuration to uvicorn. Unfortunately it seems that logs that come directly from llama_cpp get thrown to stderr and it's not configurable
(https://github.com/abetlen/llama-cpp-python/blob/ae71ad1a147b10c2c3ba99eb086521cddcc4fad4/llama_cpp/_logger.py#L30)
This change will make it so that the llm tool is installed in the workstations impage with a default logging configuration file which writes logs to /var/log/localllm.log. I don't love how different the story is if you actually run the tool directly (you have to go out of your way to get the logging) but this seems okay for now at least.
The content of the logging config is from
https://gist.github.com/liviaerxin/d320e33cbcddcc5df76dd92948e5be3b
Fixes #16