authenticate: revoke current session oauth token before sign out
After #926, we don't revoke access token before sign out anymore. It
causes sign out can not work, because right after user click on sign out
button, we redirect user to idp provider authenticate page with a valid
access token, so user is logged in immediately again.
To fix it, just revoke the access token before sign out.
Most of the flaky failure due to the fact that not all pods are ready
yet. We currently check the readiness by get all deployments and
heuristic parsing the output. So let wait for another 10 seconds before
running tests.
Currently, user's identity headers are always inserted to downstream
request. For privacy reason, it would be better to not insert these
headers by default, and let user chose whether to include these headers
per=policy basis.
Fixes#702
authorize: get claims from signed jwt
When doing databroker refactoring, all claims information were moved to
signed JWT instead of raw session JWT. But we are still looking for
claims info in raw session JWT, causes all X-Pomerium-Claim-* headers
being gone.
Fix this by looking for information from signed JWT instead.
Note that even with this fix, the X-Pomerium-Claim-Groups is still not
present, but it's another bug (see #941) and will be fixed later.
Fixes#936
* databroker: add databroker, identity manager, update cache (#864)
* databroker: add databroker, identity manager, update cache
* fix cache tests
* directory service (#885)
* directory: add google and okta
* add onelogin
* add directory provider
* initialize before sync, upate google provider, remove dead code
* add azure provider
* fix azure provider
* fix gitlab
* add gitlab test, fix azure test
* hook up okta
* remove dead code
* fix tests
* fix flaky test
* authorize: use databroker data for rego policy (#904)
* wip
* add directory provider
* initialize before sync, upate google provider, remove dead code
* fix flaky test
* update authorize to use databroker data
* implement signed jwt
* wait for session and user to appear
* fix test
* directory service (#885)
* directory: add google and okta
* add onelogin
* add directory provider
* initialize before sync, upate google provider, remove dead code
* add azure provider
* fix azure provider
* fix gitlab
* add gitlab test, fix azure test
* hook up okta
* remove dead code
* fix tests
* fix flaky test
* remove log line
* only redirect when no session id exists
* prepare rego query as part of create
* return on ctx done
* retry on disconnect for sync
* move jwt signing
* use !=
* use parent ctx for wait
* remove session state, remove logs
* rename function
* add log message
* pre-allocate slice
* use errgroup
* return nil on eof for sync
* move check
* disable timeout on gRPC requests in envoy
* fix gitlab test
* use v4 backoff
* authenticate: databroker changes (#914)
* wip
* add directory provider
* initialize before sync, upate google provider, remove dead code
* fix flaky test
* update authorize to use databroker data
* implement signed jwt
* wait for session and user to appear
* fix test
* directory service (#885)
* directory: add google and okta
* add onelogin
* add directory provider
* initialize before sync, upate google provider, remove dead code
* add azure provider
* fix azure provider
* fix gitlab
* add gitlab test, fix azure test
* hook up okta
* remove dead code
* fix tests
* fix flaky test
* remove log line
* only redirect when no session id exists
* prepare rego query as part of create
* return on ctx done
* retry on disconnect for sync
* move jwt signing
* use !=
* use parent ctx for wait
* remove session state, remove logs
* rename function
* add log message
* pre-allocate slice
* use errgroup
* return nil on eof for sync
* move check
* disable timeout on gRPC requests in envoy
* fix dashboard
* delete session on logout
* permanently delete sessions once they are marked as deleted
* remove permanent delete
* fix tests
* remove groups and refresh test
* databroker: remove dead code, rename cache url, move dashboard (#925)
* wip
* add directory provider
* initialize before sync, upate google provider, remove dead code
* fix flaky test
* update authorize to use databroker data
* implement signed jwt
* wait for session and user to appear
* fix test
* directory service (#885)
* directory: add google and okta
* add onelogin
* add directory provider
* initialize before sync, upate google provider, remove dead code
* add azure provider
* fix azure provider
* fix gitlab
* add gitlab test, fix azure test
* hook up okta
* remove dead code
* fix tests
* fix flaky test
* remove log line
* only redirect when no session id exists
* prepare rego query as part of create
* return on ctx done
* retry on disconnect for sync
* move jwt signing
* use !=
* use parent ctx for wait
* remove session state, remove logs
* rename function
* add log message
* pre-allocate slice
* use errgroup
* return nil on eof for sync
* move check
* disable timeout on gRPC requests in envoy
* fix dashboard
* delete session on logout
* permanently delete sessions once they are marked as deleted
* remove permanent delete
* fix tests
* remove cache service
* remove kv
* remove refresh docs
* remove obsolete cache docs
* add databroker url option
* cache: use memberlist to detect multiple instances
* add databroker service url
* remove cache service
* remove kv
* remove refresh docs
* remove obsolete cache docs
* add databroker url option
* cache: use memberlist to detect multiple instances
* add databroker service url
* wip
* remove groups and refresh test
* fix redirect, signout
* remove databroker client from proxy
* remove unused method
* remove user dashboard test
* handle missing session ids
* session: reject sessions with no id
* sessions: invalidate old sessions via databroker server version (#930)
* session: add a version field tied to the databroker server version that can be used to invalidate sessions
* fix tests
* add log
* authenticate: create user record immediately, call "get" directly in authorize (#931)
The log line has a well defined structure that we can process by simple
string manipulation, instead of relying on regex.
name old time/op new time/op delta
_handleLogs-12 17.3µs ±23% 1.1µs ±11% -93.81% (p=0.000 n=10+10)
name old alloc/op new alloc/op delta
_handleLogs-12 20.1kB ± 0% 4.1kB ± 0% -79.59% (p=0.002 n=8+10)
name old allocs/op new allocs/op delta
_handleLogs-12 141 ± 0% 1 ± 0% -99.29% (p=0.000 n=10+10)
handleLogs uses bufio scanner to process log output from envoy. When in
debug mode, envoy produces very long log line, causing the scanner
fails, handleLogs stop processing log. But envoy continue writing to its
stdout, which is now not consumed by any process, envoy hangs there
forever.
Fixing this by switching to use bufio.Reader instead. This is also the
real fix for failed integration test, which is interpreted wrongly by
me in #910.
After commit f62bb686d8, the default
headers, which include HSTS header will be set for response. It will
break integration test due to the interact with nginx ingress.
To fix it, set default headers without the HSTS header.
When switching to envoy, we forgot to adopt the middleware to set
response headers with options.Headers, which causes HSTS header is
missing in v0.9.0 release.
Fixes#901
Document that metricRegistry is not safe for concurrently use. While at
it, remove t.Parallel() in tests which use metricRegistry, which causes
data race, caught by:
go test -race ./internal/telemetry/metrics
config: change the default logging level to INFO
DEBUG logging level is very verbose and potentially logs sensitive data.
We should set default log level to INFO.
Updates #895Fixes#896
Caught by:
go test -race ./internal/cmd/pomerium
The ctx in Run is both read (in handle signal goroutine) and write
(when passing to errgroup context in Run), causes data race.
Fixing it, by passing the ctx to goroutine via argument instead of
accessing it directly.
