Benchmark impact on SQS ingest time & validate integration test quality on edgenet

[p0mvn]

Due to Cosmos SDK streaming API changes in v25, we had to make a change that might affect block processing time.

Currently, we have the following dashboard that tracks it:

https://grafana.osmosis.zone/d/ca8b8a4a-b38c-444c-b74a-4c8150b2b4e1/sqs-osmosis-zone?orgId=1&viewPanel=153

Test Plan

• Create an edgenet with in-place testnet
• Find out a way to track the metric from the above dashboard. Options:
  • Manually querying prometheus metrics
  • Hooking up to Grafana dashboard
  • Prints measuring time (probably easiest but please make sure it is put in the right place (see how the current metric is tracked)
• Run SQS integration tests manually against SQS connected to this testnet and see if any meaningful failures (could point out to a deeper problem)
  • This is how to run: https://github.com/osmosis-labs/sqs/blob/706a2137ef166044c16a914d908215354dfc9348/Makefile#L185
  • Please change this to http://localhost:9092 before running https://github.com/osmosis-labs/sqs/blob/706a2137ef166044c16a914d908215354dfc9348/tests/sqs_service.py#L3

DoD

• Block processing time either did not increased or 10% upper bound
• Raise concerns if any found
• Discuss a plan forward for v26 upgrade with @roman @paddy if meaningful concerns found

[czarcas7ic]

Repo created to assist in load testing https://github.com/czarcas7ic/pool-swap-load-testing

[czarcas7ic]

Leaving info here for Roman since I am unavailable next week:

adam-temp-sqs-ingest-testing is a DO node running v26 osmosis (post in place testnet and all).

This node also has pool-swap-load-testing cloned in root, and in tmux a -t 0 the validator is running

The only thing that should need to be done at this point is getting sqs to run, and then building and running the binary within pool-swap-load-testing.