When unmounting, the router entry would be tainted, preventing routing.
However, we would then unmount the router before clearing storage, so if
an error occurred the router would have forgotten the path. For auth
mounts this isn't a problem since they had a secondary check, but
regular mounts didn't (not sure why, but this is true back to at least
0.2.0). This meant you could then create a duplicate mount using the
same path which would then not conflict in the router until postUnseal.
This adds the extra check to regular mounts, and also moves the location
of the router unmount.
This also ensures that on the next router.Mount, tainted is set to the
mount entry's tainted status.
Fixes#6769
This lets other parts of Vault that can't depend on the vault package
take advantage of the subview functionality.
This also allows getting rid of BarrierStorage and vault.Entry, two
totally redundant abstractions.
* 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
* Add some requirements for versioned k/v
* Add a warning message when an upgrade is triggered
* Add path help values
* Make the kv header a const
* Add the uid to mount entry instead of options map
* Pass the backend aware uuid to the mounts and plugins
* Fix comment
* Add options to secret/auth enable and tune CLI commands (#4170)
* Switch mount/tune options to use TypeKVPairs (#4171)
* switching options to TypeKVPairs, adding bool parse for versioned flag
* flipping bool check
* Fix leases coming back from non-leased pluin kv store
* add a test for updating mount options
* Fix tests
* porting identity to OSS
* changes that glue things together
* add testing bits
* wrapped entity id
* fix mount error
* some more changes to core
* fix storagepacker tests
* fix some more tests
* fix mount tests
* fix http mount tests
* audit changes for identity
* remove upgrade structs on the oss side
* added go-memdb to vendor
This doesn't really change behavior, just what it looks like in the UX.
However, it does make tests more complicated. Most were fixed by adding
a sorting function, which is generally useful anyways.
In order to implement this efficiently, I have introduced the concept of
"singleton" backends -- currently, 'sys' and 'cubbyhole'. There isn't
much reason to allow sys to be mounted at multiple places, and there
isn't much reason you'd need multiple per-token storage areas. By
restricting it to just one, I can store that particular mount instead of
iterating through them in order to call the appropriate revoke function.
Additionally, because revocation on the backend needs to be triggered by
the token store, the token store's salt is kept in the router and
client tokens going to the cubbyhole backend are double-salted by the
router. This allows the token store to drive when revocation happens
using its salted tokens.
/cc @armon - I do a key copy within Unseal now. It tripped me up for
quite awhile that that method actually modifies the param in-place and I
can't think of any scenario that is good for the user. Do you see any
issues here?
/cc @armon - So I know the conversation we had related to this about
auth, but I think we still need to export these and do auth only at the
external API layer. If you're writing to the internal API, then all bets
are off.
The reason is simply that if you have access to the code, you can
already work around it anyways (you can disable auth or w/e), so a
compromised Vault source/binary is already a failure, and that is the
only thing that our previous unexported methods were protecting against.
If you write an external tool to access a Vault, it still needs to be
unsealed so _that_ is the primary security mechanism from an API
perspective. Once it is unsealed then the core API has full access to
the Vault, and identity/auth is only done at the external API layer, not
at the internal API layer.
The benefits of this approach is that it lets us still treat the "sys"
mount specially but at least have sys adopt helper/backend and use that
machinery and it can still be the only backend which actually has a
reference to *vault.Core to do core things (a key difference). So, an
AWS backend still will never be able to muck with things it can't, but
we're explicitly giving Sys (via struct initialization in Go itself)
a reference to *vault.Core.
