* logical/aws: Harden WAL entry creation
If AWS IAM user creation failed in any way, the WAL corresponding to the
IAM user would get left around and Vault would try to roll it back.
However, because the user never existed, the rollback failed. Thus, the
WAL would essentially get "stuck" and Vault would continually attempt to
roll it back, failing every time. A similar situation could arise if the
IAM user that Vault created got deleted out of band, or if Vault deleted
it but was unable to write the lease revocation back to storage (e.g., a
storage failure).
This attempts to harden it in two ways. One is by deleting the WAL log
entry if the IAM user creation fails. However, the WAL deletion could
still fail, and this wouldn't help where the user is deleted out of
band, so second, consider the user rolled back if the user just doesn't
exist, under certain circumstances.
Fixes#5190
* Fix segfault in expiration unit tests
TestExpiration_Tidy was passing in a leaseEntry that had a nil Secret,
which then caused a segfault as the changes to revokeEntry didn't check
whether Secret was nil; this is probably unlikely to occur in real life,
but good to be extra cautious.
* Fix potential segfault
Missed the else...
* Respond to PR feedback
* Fix for using ExplicitMaxTTL in auth method plugins.
* Reverted pb.go files for readability of PR.
* Fixed indenting of comment.
* Reverted unintended change by go test.
* Initial implemntation of returning 529 for rate limits
- bump aws iam and sts packages to v1.14.31 to get mocking interface
- promote the iam and sts clients to the aws backend struct, for mocking in tests
- this also promotes some functions to methods on the Backend struct, so
that we can use the injected client
Generating creds requires reading config/root for credentials to contact
IAM. Here we make pathConfigRoot a method on aws/backend so we can clear
the clients on successful update of config/root path. Adds a mutex to
safely clear the clients
* refactor locking and unlocking into methods on *backend
* refactor/simply the locking
* check client after grabbing lock
We do this already in the CLI, but because we do it there the UI doesn't
have ordered responses.
We could put it in the UI, but it seems like we might as well just make
it nice for all API users.
Fixes#5141
* Add request timeouts in normal request path and to expirations
* Add ability to adjust default max request duration
* Some test fixes
* Ensure tests have defaults set for max request duration
* Add context cancel checking to inmem/file
* Fix tests
* Fix tests
* Set default max request duration to basically infinity for this release for BC
* Address feedback
It's not obvious why this should be secret, and if it were considered
secret, when and what anything would ever be allowed to access it.
Likely the right way to tie secret values to particular
entities/aliases/groups would be to use the upcoming templated ACL
feature.
This change makes it so that if a lease is revoked through user action,
we set the expiration time to now and update pending, just as we do with
tokens. This allows the normal retry logic to apply in these cases as
well, instead of just erroring out immediately. The idea being that once
you tell Vault to revoke something it should keep doing its darndest to
actually make that happen.
* This changes the way policies are reported in audit logs.
Previously, only policies tied to tokens would be reported. This could
make it difficult to perform after-the-fact analysis based on both the
initial response entry and further requests. Now, the full set of
applicable policies from both the token and any derived policies from
Identity are reported.
To keep things consistent, token authentications now also return the
full set of policies in api.Secret.Auth responses, so this both makes it
easier for users to understand their actual full set, and it matches
what the audit logs now report.
Taking inspiration from
https://github.com/golang/go/issues/17604#issuecomment-256384471
suggests that taking the address of a stack variable for use in atomics
works (at least, the race detector doesn't complain) but is doing it
wrong.
The only other change is a change in Leader() detecting if HA is enabled
to fast-path out. This value never changes after NewCore, so we don't
need to grab the read lock to check it.
* Add entity information request to system view
* fixing a few comments
* sharing types between plugin and logical
* sharing types between plugin and logical
* fixing output directory for proto
* removing extra replacement
* adding mount type lookup
* empty entities return nil instead of error
* adding some comments
This takes place in two parts, since working on this exposed an issue
with response wrapping when there is a raw body set. The changes are (in
diff order):
* A CurrentWrappingLookupFunc has been added to return the current
value. This is necessary for the lookahead call since we don't want the
lookahead call to be wrapped.
* Support for unwrapping < 0.6.2 tokens via the API/CLI has been
removed, because we now have backends returning 404s with data and can't
rely on the 404 trick. These can still be read manually via
cubbyhole/response.
* KV preflight version request now ensures that its calls is not
wrapped, and restores any given function after.
* When responding with a raw body, instead of always base64-decoding a
string value and erroring on failure, on failure we assume that it
simply wasn't a base64-encoded value and use it as is.
* A test that fails on master and works now that ensures that raw body
responses that are wrapped and then unwrapped return the expected
values.
* A flag for response data that indicates to the wrapping handling that
the data contained therein is already JSON decoded (more later).
* RespondWithStatusCode now defaults to a string so that the value is
HMAC'd during audit. The function always JSON encodes the body, so
before now it was always returning []byte which would skip HMACing. We
don't know what's in the data, so this is a "better safe than sorry"
issue. If different behavior is needed, backends can always manually
populate the data instead of relying on the helper function.
* We now check unwrapped data after unwrapping to see if there were raw
flags. If so, we try to detect whether the value can be unbase64'd. The
reason is that if it can it was probably originally a []byte and
shouldn't be audit HMAC'd; if not, it was probably originally a string
and should be. In either case, we then set the value as the raw body and
hit the flag indicating that it's already been JSON decoded so not to
try again before auditing. Doing it this way ensures the right typing.
* There is now a check to see if the data coming from unwrapping is
already JSON decoded and if so the decoding is skipped before setting
the audit response.
* Allow returning list information and other data in 404s.
On read it'll output data and/or warnings on a 404 if they exist. On
list, the same behavior; the actual 'vault list' command doesn't change
behavior though in terms of output unless there are no actual keys (so
it doesn't just magically show other data).
This corrects some assumptions in response_util and wrapping.go; it also
corrects a few places in the latter where it could leak a (useless)
token in some error cases.
* Use same 404 logic in delete/put too
* Add the same secret parsing logic to the KV request functions
