* Fix xDS deadlock due to syncLoop termination.
This fixes an issue where agentless xDS streams can deadlock permanently until
a server is restarted. When this issue occurs, no new proxies are able to
successfully connect to the server.
Effectively, the trigger for this deadlock stems from the following return
statement:
https://github.com/hashicorp/consul/blob/v1.18.0/agent/proxycfg-sources/catalog/config_source.go#L199-L202
When this happens, the entire `syncLoop()` terminates and stops consuming from
the following channel:
https://github.com/hashicorp/consul/blob/v1.18.0/agent/proxycfg-sources/catalog/config_source.go#L182-L192
Which results in the `ConfigSource.cleanup()` function never receiving a
response and holding a mutex indefinitely:
https://github.com/hashicorp/consul/blob/v1.18.0/agent/proxycfg-sources/catalog/config_source.go#L241-L247
Because this mutex is shared, it effectively deadlocks the server's ability to
process new xDS streams.
----
The fix to this issue involves removing the `chan chan struct{}` used like an
RPC-over-channels pattern and replacing it with two distinct channels:
+ `stopSyncLoopCh` - indicates that the `syncLoop()` should terminate soon. +
`syncLoopDoneCh` - indicates that the `syncLoop()` has terminated.
Splitting these two concepts out and deferring a `close(syncLoopDoneCh)` in the
`syncLoop()` function ensures that the deadlock above should no longer occur.
We also now evict xDS connections of all proxies for the corresponding
`syncLoop()` whenever it encounters an irrecoverable error. This is done by
hoisting the new `syncLoopDoneCh` upwards so that it's visible to the xDS delta
processing. Prior to this fix, the behavior was to simply orphan them so they
would never receive catalog-registration or service-defaults updates.
* Add changelog.
* Refactors the leafcert package to not have a dependency on agent/consul and agent/cache to avoid import cycles. This way the xds controller can just import the leafcert package to use the leafcert manager.
The leaf cert logic in the controller:
* Sets up watches for leaf certs that are referenced in the ProxyStateTemplate (which generates the leaf certs too).
* Gets the leaf cert from the leaf cert cache
* Stores the leaf cert in the ProxyState that's pushed to xds
* For the cert watches, this PR also uses a bimapper + a thin wrapper to map leaf cert events to related ProxyStateTemplates
Since bimapper uses a resource.Reference or resource.ID to map between two resource types, I've created an internal type for a leaf certificate to use for the resource.Reference, since it's not a v2 resource.
The wrapper allows mapping events to resources (as opposed to mapping resources to resources)
The controller tests:
Unit: Ensure that we resolve leaf cert references
Lifecycle: Ensure that when the CA is updated, the leaf cert is as well
Also adds a new spiffe id type, and adds workload identity and workload identity URI to leaf certs. This is so certs are generated with the new workload identity based SPIFFE id.
* Pulls out some leaf cert test helpers into a helpers file so it
can be used in the xds controller tests.
* Wires up leaf cert manager dependency
* Support getting token from proxytracker
* Add workload identity spiffe id type to the authorize and sign functions
---------
Co-authored-by: John Murret <john.murret@hashicorp.com>
Derek Menteer