- A regression caused incomplete synchronization of the clustered Global File System 2 (GFS2). As as consequence, queries for quota limits over the network timed out. With this update, the algorithm for translating quota values to the network format has been changed to prevent indefinite cycling in the rpc.rquotad server. As a result, a file system with negative quota values can no longer make the remote procedure call quota service unresponsive.
- Previously, the reported disk usage exceeded the file system capacity because of listing disk usage on a clustered GFS2 file system when a local node was not fully synchronized. Now, disk usage and quotas are printed as signed numbers to reflect the fact that negative fluctuations in disk usage accounting do occur in unsynchronized nodes of clustered file systems. As a result, negative disk usage values are properly reported.
- Prior to this update, the rpc.quotad server terminated with the "Too many autofs mount points." error when querying for disk quotas over the network to a server that has automounted more than 64 file systems. To fix this bug, the code enumerating automounted file systems has been altered. Now, quota tools suppressing automounted file systems do not impose any limit on their number.