Questions about eos quota system


We currently so not use quotas, and were thinking to introduce them.

However, on one test instance (4.5.15 with quarkdb ns, but regularly migrated since many years) we observe that when activating quota (eos space quota default on), the quota is exceeded error is triggered even when quota is not reached (in quota ls), and even after the quota node is removed. We also have some case of MGM crashes when writing with fusex and quota is enabled.

Could it be some other parameter to consider that we forget ? Or maybe our instance is old and has some left over configuration which comes into conflict ? Some suggestion how we could troubleshoot such situation ?

Thank you