docs(pipecat-cloud): correct the readyz recycle claim and document warm instance reuse (T-3090) - #1028
Draft
jamsea wants to merge 1 commit into
Draft
docs(pipecat-cloud): correct the readyz recycle claim and document warm instance reuse (T-3090)#1028jamsea wants to merge 1 commit into
jamsea wants to merge 1 commit into
Conversation
…reuse Three related fixes found while working Daily support ticket T-3090 (unexpected memory growth on a warm Pipecat Cloud instance). 1. Health checks page. The subtitle promised "recycle instances on your own schedule", and the page read as if returning False from readyz() was a supported way to recycle an instance. It is not. Reporting not-ready does stop new sessions being routed to that instance, but nothing replaces the instance, so its process keeps running. New section says that plainly, and the subtitle no longer promises a recycle. 2. Nothing public said that a reused warm instance is the same container and the same Python process, so state a session does not release carries into the next session. New "Instance reuse and memory" section on the scaling page, plus a note on the logging page that reported CPU and memory figures are for the whole instance, not one session. 3. Capacity planning. The 5-minute cooldown sentence read as universal. Scoped it to instances above the reserved count, and said reserved agents are not cycled away by the cooldown.
Contributor
|
🔍 Mintlify preview for this branch: https://daily-docs-pcc-instance-reuse-and-readyz-t3090.mintlify.site |
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Draft. Three related Pipecat Cloud doc gaps found while working Daily support ticket T-3090 ("unexpected memory growth", a customer running
min-agents 1whose warm instance served sessions back to back for days).1. Correction: the health checks page overpromised a recycle
This one is a fix to a page that currently says something the product does not do. The subtitle promised "recycle instances on your own schedule", and the body read as if returning
Falsefromreadyz()was a supported way to recycle an instance.It is not. Reporting not-ready does stop new sessions being routed to that instance (that part of the page was right and is unchanged), but nothing then replaces the instance, so
readyz()on its own never gets you a fresh process.Changes:
2. New: a reused warm instance is the same container and the same process
Nothing public said this, and it is the single most useful missing fact for anyone debugging memory on Pipecat Cloud. The scaling page said an instance "is returned to the pool and can immediately serve another session" without saying that reuse means the same container and the same Python process, so whatever a session does not release stays for the next one.
Changes:
3. Scope fix: the cooldown sentence read as universal
The capacity planning page said an unused instance is terminated after the 5-minute cooldown. That does not hold at or below the reserved floor set by
min-agents, which is exactly the case the T-3090 customer was in. Scoped the sentence to instances above the reserved count and added a line saying reserved agents are not cycled away by the cooldown.Files
pipecat-cloud/fundamentals/health-checks.mdxpipecat-cloud/fundamentals/scaling.mdxpipecat-cloud/fundamentals/logging.mdxpipecat-cloud/guides/capacity-planning.mdxNo roadmap language anywhere: the missing recycle is stated as a current limitation only.