The "aud" claim of the logout token is supposed to contain the client-id
of the client for which the token was issued. Our current implementation of
validating that claim is somewhat broken. We only allow to configure a single
value for the allowed client id. But we have different client-ids
accessing oCIS.
This completely removes the current validation of the `aud` claim until
we come up with a working solution. As we currently require a session id
to be present in the logout token the risk not validating the `aud`
claim is pretty low.
Related: #6149
* Streamline the store implementation with and into reva
* Adapt to the cache/store refactoring in reva
* Streamline config options and their env vars
* Apply suggestions from code review
Co-authored-by: Martin <github@diemattels.at>
* Use the same database for all stores
* Bump reva
* Configure stat and filemetadata cache separately
* Fix default config
---------
Co-authored-by: Martin <github@diemattels.at>
Use access token to lookup session id. The userinfo endpoint does
not return the session id. Also add some debug logging.
Co-authored-by: Christian Richter <crichter@owncloud.com>
Co-authored-by: Michael Barz <mbarz@owncloud.com>
* refactor middleware options
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* use ocmemstore micro store implementaiton for token cache
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* refactor ocis store options, support redis sentinel
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* align cache configuration
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* database and tabe are used to build prefixes for inmemory stores
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* add global persistent store options to userlog config
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* log cache errors but continue
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* drup unnecessary type conversion
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* Better description for the default userinfo ttl
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* use global cache options for even more caches
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* don't log userinfo cache misses
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* default to stock memory store
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* use correct mem store typo string
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* split cache options, doc cleanup
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* mint and write userinfo to cache async
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* use hashed token as key
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* go mod tidy
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* update docs
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* update cache store naming
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* bring back depreceted ocis-pkg/store package for backwards compatability
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* update changelog
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* Apply suggestions from code review
Co-authored-by: kobergj <jkoberg@owncloud.com>
* revert ocis-pkg/cache to store rename
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* add waiting for each step 50 milliseconds
* starlack check
---------
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
Co-authored-by: kobergj <jkoberg@owncloud.com>
Co-authored-by: Viktor Scharf <scharf.vi@gmail.com>
Add a UserRoleAssigner implementation that extract role names from the
users' claims and creates role assignments in the settings service based
on a configured mapping of claim values to ocis role names.
Closes: #5669
This moves the lookup and the creation of the users' role assignemt out
of the user backend into its own interface. This makes the user backend
a bit simpler and allows to provide different implemenation for the user
role assignment more easily.
This removes the "withRoles" flag from the GetUserByClaims lookup and move the
functionality into a separate method. This should make the code a bit more readable
in preparation for maintaining the RoleAssignments from OIDC claims.
* add policies service
add policies proxy middleware
add policies event service
add policies grpc service
prepare ci and git environments (ci, make, readme, doc)
* add webfinger to the drone conf
* fix docs
remove not used virus scan postprocessing step
* relocate example rego file
implicitly enable and disable proxy and postprocessing policy checking by setting the query.
update configuration descriptions
* move policies
update readme
* use converter func to convert pp environment to actual environment
expose and test custom rego functions
add engine unit tests
add opa unit tests
update policies readme
Co-authored-by: Martin <github@diemattels.at>
* relocate sample policies to the deployments folder
change and document policies service port
* update index.md and small fix
* add health command
add version command
add debug server
---------
Co-authored-by: Martin <github@diemattels.at>
* use min tls 1.2
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
* add changelog
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
Signed-off-by: Jörn Friedrich Dreyer <jfd@butonic.de>
TLS for the services can be configure by setting the "OCIS_HTTP_TLS_ENABLED",
"OCIS_HTTP_TLS_CERTIFICATE" and "OCIS_HTTP_TLS_KEY" environment variables.
Currently the ocis proxy is this only service that directly accesses backend
services. It determines whether to use TLS or not by looking a the new registry
metadata "use_tls". As specific CA Cert for certificate verification
can be set with the "PROXY_HTTPS_CACERT" environment variable.
* Introduce TLS Settings for go-micro based grpc services and clients
TLS for the services can be configure by setting the OCIS_MICRO_GRPC_TLS_ENABLED"
"OCIS_MICRO_GRPC_TLS_CERTIFICATE" and "OCIS_MICRO_GRPC_TLS_KEY"
enviroment variables.
TLS for the clients can configured by setting the "OCIS_MICRO_GRPC_CLIENT_TLS_MODE"
and "OCIS_MICRO_GRPC_CLIENT_TLS_CACERT" variables.
By default TLS is disabled.
Co-authored-by: Martin <github@diemattels.at>
* Unify TLS configuration for all grpc services
All grpc service (whether they're based on reva) or go-micro use the
same set of config vars now.
TLS for the services can be configure by setting the OCIS_GRPC_TLS_ENABLED,
OCIS_GRPC_TLS_CERTIFICATE and OCIS_GRPC_TLS_KEY enviroment variables.
TLS for the clients can configured by setting the OCIS_GRPC_CLIENT_TLS_MODE
and OCIS_MICRO_GRPC_CLIENT_TLS_CACERT variables.
There are no individual per service config vars currently. If really
needed, per service tls configurations can be specified via config file.
Co-authored-by: Martin <github@diemattels.at>
Co-authored-by: Martin <github@diemattels.at>
This avoids using a public global variable. It allows us to initialize
the default client a bit later (outside of init()). That way we can e.g.
properly initialize the in-memory registry.
I refactored the proxy so that we execute the routing before the
authentication middleware. This is necessary so that we can determine
which routes are considered unprotected i.e. which routes don't need
authentication.
The old approach of the authentication middlewares had the problem that when an authenticator could not authenticate a request it would still send it to the next handler, in case that the next one can authenticate it. But if no authenticator could successfully authenticate the request, it would still be handled, which leads to unauthorized access.
Allow to switch jwt access token verification and off. Many (most?) IDP
provide JWT encoded access tokens. If ocis is configure to assume jwt
access tokens (access_token_verify_method==jwt) we now properly verify
the tokens signature and a set of standard claims ("exp", "iat" and nbf"
by way of the jwt module's standard verification and "iss" explicitliy).
This change also allows for introduction of other access token verification
mechanism in the future (e.g. through introspection (RFC7662).
The help flag is configured automatically by default already. We don't
need to redo that for every single service.
This also addresses one of the finding of "go race" (#4088)
