Updated pricing.mdx to include capacity update behavior with suspended endpoint groups #31
Conversation
robertfernandez-vast
requested review from
Colter-Downing,
abiola-vastai,
guthrie-vast and
jsbcannell
robertfernandez-vast
added
the
documentation
documentation/serverless/pricing.mdx
Outdated
| Stopping an Endpoint will pause the recruitment of GPU instances, and put the existing instances into the "Stopped" state, preventing any work from being sent to | ||
| the Endpoint group. The instances will still charge the small storage cost, but active rental and bandwidth costs will not be charged to the user account.  | ||
| Stopping an Endpoint will: | ||
| - Cause the **Serverless System** to no longer manage the GPU instances contained within the Endpoint. |
There was a problem hiding this comment.
- The Serverless Engine will no longer manage the GPU instances contained within the Endpoint.
Just so it matches above
There was a problem hiding this comment.
Will make the change. Just for my understanding, is there a difference between the Serverless system and Serverless Engine? In the Serverless overview page, the language is specifically calling out the Serverless system.
Also on the System Architecture page, it shows that Vast.ai Serverless includes:
- Serverless System
- GPU Instances
- User (Client Application)
Is there a difference between saying Serverless system and saying Serverless Engine. They appear to stated as synonymous in the documentation.
There was a problem hiding this comment.
I was using Serverless system a while ago until I realized Serverless Engine sounds cooler. All systems should be turned into engines.
documentation/serverless/pricing.mdx
Outdated
| ## Destroying GPU instances | ||
| Before destroying a GPU instance, move any data off of that GPU instance. For more help on how to do this: | ||
| - [Data Movement](https://docs.vast.ai/documentation/instances/storage/data-movement) | ||
| - [Cloud Sync](https://docs.vast.ai/documentation/instances/cloud-sync) | ||
| - [CLI Commands](https://docs.vast.ai/cli/commands) |
There was a problem hiding this comment.
We should delete this section.
There was a problem hiding this comment.
Would it be appropriate to link to the Managing Instances page? I want to make sure we do our due diligence and make it abundantly clear that it is the user's responsibility to manage inactive GPU instances and GPU instances contained within suspended Endpoint groups.
documentation/serverless/pricing.mdx
Outdated
| To destroy a GPU instance within the **serverless GUI**: | ||
| 1) Select the Endpoint group tab of the desired Endpoint group. | ||
| 2) Find the row of the desired GPU instance. | ||
| 3) Click on the **🛈** button. | ||
| 4) Click on **Destroy your instance** (trash can button). |
There was a problem hiding this comment.
We should delete this section
|
We should also think if the Pricing page is the right spot for detailing Endpoint States. |
I will talk to @guthrie-vast about this and see how to best do this. |
robertfernandez-vast
force-pushed
the
AUTO-754--Correct_Suspend_and_Stopped_endpoint_state_docs
branch
from
6a69189 to
f0011e5
Compare
robertfernandez-vast
force-pushed
the
AUTO-754--Correct_Suspend_and_Stopped_endpoint_state_docs
branch
from
f0011e5 to
74b20f0
Compare
…nds capacity updates to suspended Endpoint groups.
robertfernandez-vast
force-pushed
the
AUTO-754--Correct_Suspend_and_Stopped_endpoint_state_docs
branch
from
74b20f0 to
1493d46
Compare
There were some new implementation details regarding how Serverless interacts with suspended endpoints that now need to be captured in the documentation. The following properties are now true for suspended endpoints:
When an Endpoint is suspended:
There are also inaccuracies with the language of the Stopping an Endpoint section. I updated the language so that the stopping behavior is accurate and the billing description of inactive GPU instances is consistent with the rest of the documentation.