How is usage for quotas calculated so fast?

Question

I work on a cluster that uses a Lustre filesystem, and the administrators are able to calculate user / group usage on a daily basis. The system also uses this to calculate when users are violating their quota. So my assumption is that the total usage for a user is calculated daily, if not much more frequently.

But when I run find . -user <user> type commands, it will take over a day to traverse the filesystem, and suck down a noticeable amount of resources in the process. It seems unlikely to me that this is how the system checks for quota violation.

Does the system have a faster method? If so, is there some way I can use it?

Answer 1

The Lustre filesystem checks and updates group and user quotas on every write (with some fuzziness for speed, a "grant cache"). As a regular user you are unlikely to have read permissions for this data.