Skip to content
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.

Sign up for GitHub

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

Jump to bottom

Don't reset an active connection #3894

Merged
merged 3 commits into from Jul 14, 2020
Merged

Don't reset an active connection #3894

Changes from 1 commit
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
e9dd5b8
Don't reset an active connection
martinthomson Jul 13, 2020
c197894
Scope this more.
martinthomson Jul 14, 2020
f91ecf0
wrap
martinthomson Jul 14, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
2 changes: 1 addition & 1 deletion draft-ietf-quic-transport.md
View file
Open in desktop
Expand Up @@ -6370,7 +6370,7 @@ be arranged so that packets with a given connection ID always arrive at an
instance that has connection state, unless that connection is no longer active.

More generally, servers MUST NOT generate a stateless reset if a connection with
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

So we choose to omit saying "valid stateless reset", assuming this is implied? I.e. a server may generate a stateless reset for a connection handled by a different server, but it is ok if they do not share stateless reset secret, since the resulting stateless reset will be invalid.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, I think that "valid" is implied. In the case of one server generating a stateless reset based on a connection ID that came from another server, that wouldn't affect that other connection. It would only affect the connection that the server generating the stateless reset might have created, if there was such a connection.

Copy link
Contributor

@igorlord igorlord Jul 13, 2020 •
edited

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ok. Just to be clear, it is not about servers that generate connection IDs. It is about sets of servers that share the same Stateless Reset secrets. Servers MUST NOT send a stateless reset for a connection ID, if it is possible that this connection ID is active on any server that is still able to handle this connection and shares the same Stateless Reset secret.

Note: this implies that backend servers behind a 5-tuple based load balancer cannot share a stateless reset secret, or they must identify the backend server from the connection ID and not respond with a stateless reset for a connection ID generated by a different server, if that server may still be operational.

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes (with the exception noted in the paragraph that follows, which won't affect your 5-tuple-based load balancer).

Copy link
Contributor

@igorlord igorlord Jul 13, 2020 •
edited

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The following paragraph is a bit misleading in case of 5-tuple load balancers. It is talking about routing of "the packet". The specific packet might have a fixed routing due to its 5-tuple. An attacker can rewrite the packet's source address to replace it with its own, causing the forged packet to be routed to a different server. What matters is a possibility that routing of packets with this CID -- can their routing be affected by the attacker?

(In my formulation above, "still able to handle this connection" included "can still have packets containing this connection ID routed to it")

the corresponding connection ID could be active.
the corresponding connection ID could be active on any endpoint using the same static key.

In the case of a cluster that uses dynamic load balancing, it's possible that a
change in load balancer configuration could occur while an active instance
Expand Down