v0.3.10

Release Indexfiy 3.10 (#1670)

v0.3.9

Bump Server version to release new image version

This would fix tensorlake CI

v0.3.8

Update version (#1648)

* Add support for JSON and SSE responses to invoke endpoint

* update tensorlake version, indexify server and executor version

* update poetry lock

* fix import

* fix compilation errors

* fix test

v0.3.7

Bump axios in /server/ui in the npm_and_yarn group across 1 directory…

v0.3.6

Bump indexify-server version (#1629)

v0.3.5

Bump version to release indexify

v0.3.4

Updated event attributes, used spans where it seemed sensible. (#1489)

* Updated event attributes, used spans where it seemed sensible.

* Use consistent logging on executor

* Address naming inconsistencies.

* Print 1 message per fn_executor action.

* Rework fn_executor messages.

* Log executor_id and log inline while we are removing executors

* Lint for the linter god.

* Reimplemented logger changes, hopefully properly this time.

* Standardize on fn instead of fn_name

* Removed redundant message

* Added missing fn renames

* Lint for the linter god

* Fix broken merge

* Force use latest tensorlake.

* Minor fixes in logging tags + tensorlake update

---------

Co-authored-by: Julio Martinez <[email protected]>
Co-authored-by: Eugene Batalov <[email protected]>

v0.3.3

Install indexify into images build using indexify-cli

OSS Indexify users need indexify to be installed in their function
images. The latest Tensorlake SDK is not installing them anymore.

v0.3.2

Introduce FunctionExecutorResources into Server resource model

FunctionExecutorResources represent resources allocated for FE
while the previously used NodeResources only represents resources
configured for a graph node. The key difference between the two
is that NodeResources includes all possible GPU model:count pairs
for a function while FunctionExecutorResources represent particular
GPU model:count (and other resources) allocated from a particular
Executor that has enough free GPU model:count.

This change also aligns with FunctionExecutorResources object
used in Executor proto API.

Update Python version to 3.10

This is the new min supported version in Tensorlake SDK.

Add unit tests for HostResources data_model class

They aim to cover all main resource accounting scenarios to make
sure we don't have coverage gaps.

v0.3.1

Removed whitespace and typos