TTL: 1 second difference

[TVenuMadhav]

If we set the configuration ttl.access_token as 1h , the token.expiry_at should be 3600 but comes to be 3599. There is a 1 second difference. (Same for all other configurations like 2h, 10h etc)

Ideally, this shouldn't be the case, is there any reason for this?

[vinckr]

Can you provide some more info on your setup, how are you running hydra, what version, in what environment, with what configuration?
I agree that the lifetime should be exact.
How did you determine that it is off by one second? Can you give me some pointers on how to reproduce maybe?

[TVenuMadhav]

Also are you using MySQL?

postgres

[TVenuMadhav]

What environment are you using this one? If you have little CPU or very high BCrypt work factor this could potentially happen

tested on
-> local (docker compose container based [details provided at top of the thread]), and
-> staging environment(s) (decent CPU on staging)

[aeneasr]

We could probably only fix that by specifying the time once and then re-using it. But I'm wondering if it is worth the effort.

[TVenuMadhav]

if this would be fixed in the future, any timelines that we can be informed about?

[TVenuMadhav]

Opened an issue, so this bug can be tracked -> #2507