-
Notifications
You must be signed in to change notification settings - Fork 1
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Registry-sweeper upgrade for multitenant registry #120
Comments
Alex is making progress on this ticket. |
Status: Sagar to take a look at necessary roles to perform this action. |
Currently being tested on MCP - image has been pushed to ECR but need ECS settings to continue. Will follow up with @sjoshi-jpl tomorrow Backwards compatibility with non-MT registry has been manually tested. |
Deployed in dev but issues with 403 errors. |
status: auth is handled, but lidvids appear to be typed as |
@alexdunnjpl is unblocked on this ticket and will resume the testing when he is done with the resolution of the data migration. |
Status: nontrivial refactoring required. Need to look at EC2 scratch scripts in Separate credential instantiation from auth/opensearch-py client instantiation. Will be straightforward, but if I don't write it down now I'll need to figure it out again later. |
So more work is needed. |
Unsure about current state/status - probably will need to cherrypick/reimplement PR from new branch |
@jordanpadams pulling from sprint for now - but if this is a high priority, feel free to add it back. I've lost track of where this sits in terms of priority and being blocked by other work. |
Cherrypick complete(?) and open in #130, but not tested yet |
Status: blocked in testing this on prod. |
Status: IAM role created, but looks like permissions need tweaking. Working with @sjoshi-jpl to resolve this. |
@jordanpadams @tloubrieu-jpl @sjoshi-jpl Is the ECS sweepers deployment in another ticket? If not, feel free to re-open this if you don't want to create a separate one. |
@jordanpadams @tloubrieu-jpl psa ancestry sweeper terminated on mcp prod1 due to out-of-memory condition. All other sweepers/nodes have completed successfully. Recommend against attempting to re-run on prod1 - it can wait until we have it deployed to ECS (where we can size accordingly). |
💡 Description
Using opensearch serverless
We now have one single opensearch serverless URL.
Sweeper still takes a single node as argument, AWS infrastructure takes care of running the needed sweepers (one for each node). There will be one task definition per node.
Create new roles read/write access that will be associated to the sweeper's ECS task.
We should not need to signed the HTTP requests to connect to OpenSearch and to do it we should use a AWS SDK (see https://docs.aws.amazon.com/opensearch-service/latest/developerguide/serverless-sdk.html). @sjoshi-jpl has example of using this code.
If we want to run sweeper from a local laptop we would still need the signed URLs to be implemented.
⚔️ Parent Epic / Related Tickets
No response
The text was updated successfully, but these errors were encountered: