The session token renewal does
1) Read the old token
2) Write a new token
3) Delete the old token
If two processes succeed to read the old token there can be two new tokens because
the queries were not run in a transaction. This is particularly problematic on
clustered DBs where 1) would go to a read node and 2) and 3) go to a write node.
Signed-off-by: Christoph Wurst <christoph@winzerhof-wurst.at>
This is an helpful helper that should be used in more place than just
server and this is already the case with groupfodlers, deck, user_oidc
and more using it, so let's make it public
Signed-off-by: Carl Schwan <carl@carlschwan.eu>
For passwords bigger than 250 characters, use a bigger key since the
performance impact is minor (around one second to encrypt the password).
For passwords bigger than 470 characters, give up earlier and throw
exeception recommanding admin to either enable the previously enabled
configuration or use smaller passwords.
Signed-off-by: Carl Schwan <carl@carlschwan.eu>
This adds an option to disable storing passwords in the database. This
might be desirable when using single use token as passwords or very
large passwords.
Signed-off-by: Carl Schwan <carl@carlschwan.eu>
otherwise, when using object store, we loose track of which files the user owns before we can delete them
Signed-off-by: Robin Appelman <robin@icewind.nl>
The auth token activity logic works as follows
* Read auth token
* Compare last activity time stamp to current time
* Update auth token activity if it's older than x seconds
This works fine in isolation but with concurrency that means that
occasionally the same token is read simultaneously by two processes and
both of these processes will trigger an update of the same row.
Affectively the second update doesn't add much value. It might set the
time stamp to the exact same time stamp or one a few seconds later. But
the last activity is no precise science, we don't need this accuracy.
This patch changes the UPDATE query to include the expected value in a
comparison with the current data. This results in an affected row when
the data in the DB still has an old time stamp, but won't affect a row
if the time stamp is (nearly) up to date.
This is a micro optimization and will possibly not show any significant
performance improvement. Yet in setups with a DB cluster it means that
the write node has to send fewer changes to the read nodes due to the
lower number of actual changes.
Signed-off-by: Christoph Wurst <christoph@winzerhof-wurst.at>
For historic reasons we couldn't add a nullable type hint before
nullable type hints were supported by our target php versions. This is
now possible.
Signed-off-by: Christoph Wurst <christoph@winzerhof-wurst.at>
Else you can end up that you renewed your password (LDAP for example).
But they still don't work because you did not use them before you logged
in.
Signed-off-by: Roeland Jago Douma <roeland@famdouma.nl>
The IConfig service is documented to handle its data as strings, hence
this changes the code a bit to ensure we store keys as string and
convert them back when reading.
Signed-off-by: Christoph Wurst <christoph@winzerhof-wurst.at>
On some systems with a lot of users this creates a lot of extra DB
writes.
Being able to increase this interval helps there.
Signed-off-by: Roeland Jago Douma <roeland@famdouma.nl>
