[rfe] rate-limit metadata API
Bug #1989199 reported by
Guillaume Espanel
This bug affects 2 people
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
neutron |
Fix Committed
|
Wishlist
|
Miguel Lavalle |
Bug Description
At the moment, there is no limit on how many requests the metadata-agent
will handle. Some users may sometimes run scripts in their instances that
try to query the metadata endpoint at high rate (for example a bugged k8s
cloud controller manager), causing an increased load on some or all the
component above the metadata-agent.
We'd like to add some rate-limiting around the metadata-agent level to give
some protection to the other components. We were thinking to implement
that through a change to the metadata's haproxy's configuration using
stick-tables. In that case, the rate-limited queries would get a 429
without even being passed to the metadata-agent itself.
Changed in neutron: | |
status: | New → In Progress |
importance: | Undecided → Wishlist |
Changed in neutron: | |
assignee: | nobody → Miguel Lavalle (minsel) |
status: | In Progress → Fix Committed |
To post a comment you must log in.
We discussed this RFE during our drivers meeting, see the logs: /meetings. opendev. org/meetings/ neutron_ drivers/ 2022/neutron_ drivers. 2022-09- 09-14.00. log.html# l-51
https:/
The agreement was to approve this RFE, and ask for a spec where the details can be discussed. thanks for proposing.