feat(bigtable): add preemptive connection recycler#13860
feat(bigtable): add preemptive connection recycler#13860sushanb merged 6 commits intogoogleapis:mainfrom
Conversation
There was a problem hiding this comment.
Code Review
This pull request introduces a preemptive connection recycler to manage the lifecycle of connections in the Bigtable client's connection pool, including configuration, background logic, and client integration. However, a security vulnerability was identified: the random number generator used for connection recycling jitter is seeded with a predictable value, which could lead to synchronized connection recycling at scale and impact service availability. It is recommended to use a cryptographically secure random source for seeding. Additionally, there are suggestions to improve code clarity, conventions, and configuration handling, such as addressing a non-idiomatic package name, a silent override of user configuration, and minor typos.
| EnableDynamicChannelPool bool | ||
| // DisableDynamicChannelPool disables the dynamic channel resizing based on load | ||
| // Dynamic channel resizing is enabled by default to resize based on load and avoid queuing of requests. | ||
| DisableDynamicChannelPool bool |
There was a problem hiding this comment.
This makes the code a little hard to read. Is it possible to have a constructor function that returns true for these variables by default?
No description provided.