[feat] add observability stack to MP mode#2638
Conversation
Signed-off-by: royyhuang <roy.y.huang@gmail.com>
Signed-off-by: royyhuang <roy.y.huang@gmail.com>
Summary of ChangesHello @royyhuang, I'm Gemini Code Assist1! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed! This pull request integrates an observability stack into the multi-process (MP) mode, specifically targeting the distributed storage components. It establishes a clear listener pattern for various storage events, enabling the collection of metrics and insights into cache operations. The changes lay the groundwork for monitoring the performance and behavior of the L1 cache and overall storage manager interactions by providing granular event notifications and a dedicated Prometheus controller. Highlights
🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console. Changelog
Activity
Using Gemini Code AssistThe full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips. Invoking Gemini You can request assistance from Gemini at any point by creating a comment using either
Customization To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a Limitations & Feedback Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for Github and other Google products, sign up here. You can also get AI-powered code generation, chat, as well as code reviews directly in the IDE at no cost with the Gemini Code Assist IDE Extension. Footnotes
|
There was a problem hiding this comment.
Code Review
This pull request introduces an observability stack for the multi-process mode by adding a listener pattern to track storage events. The changes are well-structured, introducing StorageManagerListener and updating related components. However, there are a few critical issues in the implementation of the new listener interfaces. Specifically, StorageStatsListener is missing implementations for abstract methods, and EvictionPolicy has incorrect method signatures for the new listener methods, both of which will cause runtime errors. There are also some minor issues with a misleading comment and an incomplete loop in the new PrometheusController.
Signed-off-by: royyhuang <roy.y.huang@gmail.com>
Signed-off-by: royyhuang <roy.y.huang@gmail.com>
Signed-off-by: royyhuang <roy.y.huang@gmail.com>
Signed-off-by: royyhuang <roy.y.huang@gmail.com>
Signed-off-by: royyhuang <roy.y.huang@gmail.com>
Signed-off-by: royyhuang <roy.y.huang@gmail.com>
There was a problem hiding this comment.
LGTM. Some future tasks:
- Move the observability module to something like
lmcache/v1/mp_observability(and the configuration). - Update the Prometheus controller implementation to enable a more flexible way to initialize the loggers (see the detail below)
- More stats loggers for L2, mp servers, vLLM integrations
| def __init__( | ||
| self, | ||
| storage_manager: "StorageManager", | ||
| l1_manager: L1Manager, | ||
| log_interval: float, | ||
| ): | ||
| super().__init__(storage_manager, l1_manager) | ||
|
|
||
| self._log_interval = log_interval | ||
| self.all_loggers: List[PrometheusLogger] = [] | ||
|
|
||
| self.sm_stats_logger: StorageManagerStatsLogger = StorageManagerStatsLogger() | ||
| self.get_storage_manager().register_listener(self.sm_stats_logger) | ||
| self.all_loggers.append(self.sm_stats_logger) | ||
|
|
||
| self.l1_stats_logger: L1ManagerStatsLogger = L1ManagerStatsLogger() | ||
| self.get_l1_manager().register_listener(self.l1_stats_logger) | ||
| self.all_loggers.append(self.l1_stats_logger) |
There was a problem hiding this comment.
In the future, we can let each module (L1Manager, L2Manager, etc) to intialize their own loggers, and register those loggers into the Prometheus controller.
The Prometheus controller should be a singleton, and can be accessed by other components for logger registration.
In this case, the Prometheus controller itself does not storage_manager and l1_manager during intialization
* [feat] adding storage manager listener and observability skeleton Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove local uv.lock and coding agent markdown Signed-off-by: royyhuang <roy.y.huang@gmail.com> * [feat] add prometheus logger interface and prometheus controller Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove misleading no ops in eviction policy Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix code quality check and prometheus naming conflicts Signed-off-by: royyhuang <roy.y.huang@gmail.com> * refactor storage manager stats logger into stats logger per storage tier Signed-off-by: royyhuang <roy.y.huang@gmail.com> * address comments Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix tests Signed-off-by: royyhuang <roy.y.huang@gmail.com> --------- Signed-off-by: royyhuang <roy.y.huang@gmail.com>
* [feat] adding storage manager listener and observability skeleton Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove local uv.lock and coding agent markdown Signed-off-by: royyhuang <roy.y.huang@gmail.com> * [feat] add prometheus logger interface and prometheus controller Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove misleading no ops in eviction policy Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix code quality check and prometheus naming conflicts Signed-off-by: royyhuang <roy.y.huang@gmail.com> * refactor storage manager stats logger into stats logger per storage tier Signed-off-by: royyhuang <roy.y.huang@gmail.com> * address comments Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix tests Signed-off-by: royyhuang <roy.y.huang@gmail.com> --------- Signed-off-by: royyhuang <roy.y.huang@gmail.com>
* [feat] adding storage manager listener and observability skeleton Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove local uv.lock and coding agent markdown Signed-off-by: royyhuang <roy.y.huang@gmail.com> * [feat] add prometheus logger interface and prometheus controller Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove misleading no ops in eviction policy Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix code quality check and prometheus naming conflicts Signed-off-by: royyhuang <roy.y.huang@gmail.com> * refactor storage manager stats logger into stats logger per storage tier Signed-off-by: royyhuang <roy.y.huang@gmail.com> * address comments Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix tests Signed-off-by: royyhuang <roy.y.huang@gmail.com> --------- Signed-off-by: royyhuang <roy.y.huang@gmail.com> Signed-off-by: Ofer Kiselov Nahman <ofer.kiselovnahman@weka.io>
* [feat] adding storage manager listener and observability skeleton Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove local uv.lock and coding agent markdown Signed-off-by: royyhuang <roy.y.huang@gmail.com> * [feat] add prometheus logger interface and prometheus controller Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove misleading no ops in eviction policy Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix code quality check and prometheus naming conflicts Signed-off-by: royyhuang <roy.y.huang@gmail.com> * refactor storage manager stats logger into stats logger per storage tier Signed-off-by: royyhuang <roy.y.huang@gmail.com> * address comments Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix tests Signed-off-by: royyhuang <roy.y.huang@gmail.com> --------- Signed-off-by: royyhuang <roy.y.huang@gmail.com>
* [feat] adding storage manager listener and observability skeleton Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove local uv.lock and coding agent markdown Signed-off-by: royyhuang <roy.y.huang@gmail.com> * [feat] add prometheus logger interface and prometheus controller Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove misleading no ops in eviction policy Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix code quality check and prometheus naming conflicts Signed-off-by: royyhuang <roy.y.huang@gmail.com> * refactor storage manager stats logger into stats logger per storage tier Signed-off-by: royyhuang <roy.y.huang@gmail.com> * address comments Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix tests Signed-off-by: royyhuang <roy.y.huang@gmail.com> --------- Signed-off-by: royyhuang <roy.y.huang@gmail.com>
* [feat] adding storage manager listener and observability skeleton Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove local uv.lock and coding agent markdown Signed-off-by: royyhuang <roy.y.huang@gmail.com> * [feat] add prometheus logger interface and prometheus controller Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove misleading no ops in eviction policy Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix code quality check and prometheus naming conflicts Signed-off-by: royyhuang <roy.y.huang@gmail.com> * refactor storage manager stats logger into stats logger per storage tier Signed-off-by: royyhuang <roy.y.huang@gmail.com> * address comments Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix tests Signed-off-by: royyhuang <roy.y.huang@gmail.com> --------- Signed-off-by: royyhuang <roy.y.huang@gmail.com> Signed-off-by: shaoxiawjc <wjc2800@163.com>
* [feat] adding storage manager listener and observability skeleton Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove local uv.lock and coding agent markdown Signed-off-by: royyhuang <roy.y.huang@gmail.com> * [feat] add prometheus logger interface and prometheus controller Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove misleading no ops in eviction policy Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix code quality check and prometheus naming conflicts Signed-off-by: royyhuang <roy.y.huang@gmail.com> * refactor storage manager stats logger into stats logger per storage tier Signed-off-by: royyhuang <roy.y.huang@gmail.com> * address comments Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix tests Signed-off-by: royyhuang <roy.y.huang@gmail.com> --------- Signed-off-by: royyhuang <roy.y.huang@gmail.com> Signed-off-by: Aaron Wu <aaron.wu@dell.com>
* [feat] adding storage manager listener and observability skeleton Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove local uv.lock and coding agent markdown Signed-off-by: royyhuang <roy.y.huang@gmail.com> * [feat] add prometheus logger interface and prometheus controller Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove misleading no ops in eviction policy Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix code quality check and prometheus naming conflicts Signed-off-by: royyhuang <roy.y.huang@gmail.com> * refactor storage manager stats logger into stats logger per storage tier Signed-off-by: royyhuang <roy.y.huang@gmail.com> * address comments Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix tests Signed-off-by: royyhuang <roy.y.huang@gmail.com> --------- Signed-off-by: royyhuang <roy.y.huang@gmail.com>
* [feat] adding storage manager listener and observability skeleton Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove local uv.lock and coding agent markdown Signed-off-by: royyhuang <roy.y.huang@gmail.com> * [feat] add prometheus logger interface and prometheus controller Signed-off-by: royyhuang <roy.y.huang@gmail.com> * remove misleading no ops in eviction policy Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix code quality check and prometheus naming conflicts Signed-off-by: royyhuang <roy.y.huang@gmail.com> * refactor storage manager stats logger into stats logger per storage tier Signed-off-by: royyhuang <roy.y.huang@gmail.com> * address comments Signed-off-by: royyhuang <roy.y.huang@gmail.com> * fix tests Signed-off-by: royyhuang <roy.y.huang@gmail.com> --------- Signed-off-by: royyhuang <roy.y.huang@gmail.com>
This PR introduces observability stack for our multiprocess mode via newly introduced EventListener class.
Special notes for your reviewers:
If applicable: