fix/enhancement - Fix slow auth tokens. #164
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
jwag956
force-pushed
the
tokenhash
branch
2 times, most recently
from
97cadd5 to
22eb548
Compare
Current auth tokens are slow because they use the user's password (hashed) as a uniquifier (the user id isn't really enough since it might be reused). This requires checking the (hashed) password against what is in the token on EVERY request - however hashing is (on purpose) slow. So this can add almost a whole second to every request! This PR introduces a new UserModel field - fs_uniquifier - that if present in the UserModel will be populated and used rather than the password. This results in 50x reduction in time when authenticating via token. Furthermore, the actual token verification has been moved from request_loader into the UserMixin - so that it could be overridden (creating the auth_token already was in the UserMixin). Note that this does require a DB migration to add the field. The fsqla model has been updated, and docs describing at least one way to migrate the DB have been added. 2 new backwards compatibility configurations have been added that can revert some new default behavior. First - in the past- the auth token was included always in JSON responses to login, reset and change - even if the caller was a browser. This is really not great since auth tokens may have very long expire times (or none) and it shouldn't even be sent if not needed. Now, by default, the auth token is NEVER returned - the caller may request is during login, reset, or change by adding the 'include_auth_token' query param. Second, since auth tokens used to be checked against the hashed password - changing a user's password meant that any outstanding auth tokens would be invalidated. That seems like strange behavior - so by default, tokens that are verified with the new fs_uniquifier won't be invalidated just because the user's password changes. The BACKWARDS_COMPAT_AUTH_TOKEN_INVALID config variable will cause the fs_uniquifier to be changed whenever the user's password changes, thus restoring the older behavior. closes: #156
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Current auth tokens are slow because they use the user's password (hashed) as a uniquifier (the
user id isn't really enough since it might be reused). This requires checking the (hashed) password against
what is in the token on EVERY request - however hashing is (on purpose) slow. So this can add almost a whole second
to every request!
This PR introduces a new UserModel field - fs_uniquifier - that if present in the UserModel will be populated and used rather than the password. This results in 50x reduction in time when authenticating via token.
Furthermore, the actual token verification has been moved from request_loader into the UserMixin - so
that it could be overridden (creating the auth_token already was in the UserMixin).
Note that this does require a DB migration to add the field. The fsqla model has been updated, and docs
describing at least one way to migrate the DB have been added.
2 new backwards compatibility configurations have been added that can revert some new default behavior.
First - in the past- the auth token was included always in JSON responses to login, reset and change -
even if the caller was a browser. This is
really not great since auth tokens may have very long expire times (or none) and it shouldn't even be sent if not needed.
Now, by default, the auth token is NEVER returned - the caller may request is during login, reset, or change by adding the
'include_auth_token' query param.
Second, since auth tokens used to be checked against the hashed password - changing a user's password meant that any
outstanding auth tokens would be invalidated. That seems like strange behavior - so by default, tokens that are verified
with the new fs_uniquifier won't be invalidated just because the user's password changes. The
BACKWARDS_COMPAT_AUTH_TOKEN_INVALID config variable will cause the fs_uniquifier to be changed whenever the user's password changes, thus restoring the older behavior.
closes: #156