There are two restrictions:
- Writes from the primary DC which explicitly target a secondary DC.
- Writes to a secondary DC that do not explicitly target the primary DC.
The first restriction is because the config entry is not supported in
secondary datacenters.
The second restriction is to prevent the scenario where a user writes
the config entry to a secondary DC, the write gets forwarded to the
primary, but then the config entry does not apply in the secondary.
This makes the scope more explicit.
The duo of `makeUpstreamFilterChainForDiscoveryChain` and `makeListenerForDiscoveryChain` were really hard to reason about, and led to concealing a bug in their branching logic. There were several issues here:
- They tried to accomplish too much: determining filter name, cluster name, and whether RDS should be used.
- They embedded logic to handle significantly different kinds of upstream listeners (passthrough, prepared query, typical services, and catch-all)
- They needed to coalesce different data sources (Upstream and CompiledDiscoveryChain)
Rather than handling all of those tasks inside of these functions, this PR pulls out the RDS/clusterName/filterName logic.
This refactor also fixed a bug with the handling of [UpstreamDefaults](https://www.consul.io/docs/connect/config-entries/service-defaults#defaults). These defaults get stored as UpstreamConfig in the proxy snapshot with a DestinationName of "*", since they apply to all upstreams. However, this wildcard destination name must not be used when creating the name of the associated upstream cluster. The coalescing logic in the original functions here was in some situations creating clusters with a `*.` prefix, which is not a valid destination.
* ui: Filter global intentions list by namespace and partition
Filters global intention listing by the current partition rather than trying to use a wildcard.
Fixes an issue described in #10132, where if two DCs are WAN federated
over mesh gateways, and the gateway in the non-primary DC is terminated
and receives a new IP address (as is commonly the case when running them
on ephemeral compute instances) the primary DC is unable to re-establish
its connection until the agent running on its own gateway is restarted.
This was happening because we always preferred gateways discovered by
the `Internal.ServiceDump` RPC (which would fail because there's no way
to dial the remote DC) over those discovered in the federation state,
which is replicated as long as the primary DC's gateway is reachable.
Currently getCARoots could return an empty object with an empty trust
domain before the CA is initialized. This commit returns an error while
there is no CA config or no trust domain.
There could be a CA config and no trust domain because the CA config can
be created in InitializeCA before initialization succeeds.
* state: port KV and Tombstone tables to new pattern
* go fmt'ed
* handle wildcards for tombstones
* Fix graveyard ent vs oss
* fix oss compilation error
* add partition to tombstones and kv state store indexes
* refactor to use `indexWithEnterpriseIndexable`
* Apply suggestions from code review
Co-authored-by: Chris S. Kim <ckim@hashicorp.com>
Co-authored-by: R.B. Boyer <4903+rboyer@users.noreply.github.com>
* add `singleValueID` implementation assertions
* partition `tableSessions` table
* fix sessions to use UUID and fix prefix index
* fix oss build
* clean up unused functions
* fix oss compilation
* add a partition indexer for sessions
* Fix oss to not have partition index
* fix oss tests
* remove unused func `prefixIndexFromServiceNameAsString`
* fix test error check
* remove unused operations_ent.go and operations_oss.go func
* remove unused const
Co-authored-by: Daniel Nephin <dnephin@hashicorp.com>
Co-authored-by: Chris S. Kim <ckim@hashicorp.com>
Co-authored-by: R.B. Boyer <4903+rboyer@users.noreply.github.com>
* state: port KV and Tombstone tables to new pattern
* go fmt'ed
* handle wildcards for tombstones
* Fix graveyard ent vs oss
* fix oss compilation error
* add partition to tombstones and kv state store indexes
* refactor to use `indexWithEnterpriseIndexable`
* partition kvs indexID table
* add `partitionedIndexEntryName` in oss for test purpose
* Apply suggestions from code review
Co-authored-by: Chris S. Kim <ckim@hashicorp.com>
Co-authored-by: R.B. Boyer <4903+rboyer@users.noreply.github.com>
* add `singleValueID` implementation assertions
* remove entmeta reference from oss
Co-authored-by: R.B. Boyer <4903+rboyer@users.noreply.github.com>
Co-authored-by: Daniel Nephin <dnephin@hashicorp.com>
Co-authored-by: Chris S. Kim <ckim@hashicorp.com>
Co-authored-by: R.B. Boyer <4903+rboyer@users.noreply.github.com>
Previously secondaryInitialize would return nil in this case, which prevented the
deferred initialize from happening, and left the CA in an uninitialized state until a config
update or root rotation.
To fix this I extracted the common parts into the delegate implementation. However looking at this
again, it seems like the handling in secondaryUpdateRoots is impossible, because that function
should never be called before the secondary is initialzied. I beleive we can remove some of that
logic in a follow up.
These two fields do not appear to be used anywhere. We use the structs.ACLPolicy ID in the
ACLResolver cache, but the acl.Policy ID and revision are not used.
* Support Vault Namespaces explicitly in CA config
If there is a Namespace entry included in the Vault CA configuration,
set it as the Vault Namespace on the Vault client
Currently the only way to support Vault namespaces in the Consul CA
config is by doing one of the following:
1) Set the VAULT_NAMESPACE environment variable which will be picked up
by the Vault API client
2) Prefix all Vault paths with the namespace
Neither of these are super pleasant. The first requires direct access
and modification to the Consul runtime environment. It's possible and
expected, not super pleasant.
The second requires more indepth knowledge of Vault and how it uses
Namespaces and could be confusing for anyone without that context. It
also infers that it is not supported
* Add changelog
* Remove fmt.Fprint calls
* Make comment clearer
* Add next consul version to website docs
* Add new test for default configuration
* go mod tidy
* Add skip if vault not present
* Tweak changelog text
* Remove some usage of md5 from the system
OSS side of https://github.com/hashicorp/consul-enterprise/pull/1253
This is a potential security issue because an attacker could conceivably manipulate inputs to cause persistence files to collide, effectively deleting the persistence file for one of the colliding elements.
Signed-off-by: Mark Anderson <manderson@hashicorp.com>
Port of: Ensure we check intention service prefix permissions for per service (#11270)
Previously, when showing some action buttons for 'per service intentions' we used a global 'can I do something with any intention' permission to decide whether to show a certain button or not. If a user has a token that does not have 'global' intention permissions, but does have intention permissions on one or more specific services (for example via service / service_prefix), this meant that we did not show them certain buttons required to create/edit the intentions for this specific service.
This PR adds that extra permissions check so we now check the intentions permissions per service instead of using the 'global' "can I edit intentions" question/request.
**Notes:**
- If a HTML button is `disabled` this means tippy.js doesn't adopt the
popover properly and subsequently hide it from the user, so aswell as
just disabling the button so you can't active the popover, we also don't
even put the popover on the page
- If `ability.item` or `ability.item.Resources` are empty then assume no access
**We don't try to disable service > right hand side intention actions here**
Whether you can create intentions for a service depends on the
_destination_ of the intention you would like to create. For the
topology view going from the LHS to the center, this is straightforwards
as we only need to know the permissions for the central service, as when
you are going from the LHS to the center, the center is the
_destination_.
When going from the center to the RHS the _destination[s]_ are on the
RHS. This means we need to know the permissions for potentially 1000s of
services all in one go in order to know when to show a button or not.
We can't realistically discover the permissions for service > RHS
services as we'd have either make a HTTP request per right hand service,
or potentially make an incredibly large POST request for all the
potentially 1000s of services on the right hand side (more preferable to
1000s of HTTP requests).
Therefore for the moment at least we keep the old functionality (thin client)
for the middle to RHS here. If you do go to click on the button and you
don't have permissions to update the intention you will still not be
able to update it, only you won't know this until you click the button
(at which point you'll get a UI visible 403 error)
Note: We reversed the conditional here between 1.10 and 1.11
So this make 100% sense that the port is different here to 1.11
* docs: consul-k8s uninstall with namespace
Uninstall with namespace
* change release name to consul in uninstall
* Update website/content/docs/k8s/operations/uninstall.mdx
Co-authored-by: trujillo-adam <47586768+trujillo-adam@users.noreply.github.com>
* add --create-namespace command to install for custom values file
Co-authored-by: trujillo-adam <47586768+trujillo-adam@users.noreply.github.com>
trujillo-adam
freddygv
Freddy
mrspanishviking
99
John Cowen
Kyle Havlovitz
Mark Anderson
Daniel Upton
Dhia Ayachi
Giulio Micheloni
David Yu
Daniel Nephin
R.B. Boyer
Evan Culver
Luke Kysow
Kenia
Connor
FFMMM
Kevin Wang
Kyle Rarey