Hi. I have a funny zfs dataset on a recent stable, which listing takes several minutes. In the same time - it runs last available pool version with all available feature flags enabled (however, problem manifested itself on a one year old stable) - it's an old /tmp directory - it used to hold 2.5 millions of files (operational error from nagios) - now it's empty - zpool status shows no errors - scrub found no errors - ktrace shows it's getdirentries() that is waiting for something - ps shows zio->io_(trimmed)wchan state Does someone take a look into this, in case this is some kind of rare bug (and I suppose it is) ? If not, I will just delete it - I've already recreated /tmp and renamed old one. Eugene.