While RAID logical volumes and thinly provisioned logical volumes can be used in a cluster when activated exclusively, there is currently no support for thin provisioning on top of RAID in a cluster. This is the case even if the combination is activated exclusively. Currently this combination is only supported in LVM's single machine non-clustered mode.

If a thin-pool is filled to capacity, it may be possible to lose some writes even if the pool is being grown at that time. This is because a resize operation (even an automated one) will attempt to flush outstanding I/O to the storage device prior to the resize being performed. Since there is no room in the thin-pool, the I/O operations must be errored first to allow the grow to succeed. Once the thin pool has grown, the logical volumes associated with the thin-pool will return to normal operation.

As a workaround to this problem, set 'thin_pool_autoextend_threshold' and 'thin_pool_autoextend_percent' appropriately for your needs in the lvm.conf file. Do not set the threshold so high or the percent so low that your thin-pool will reach full capacity so quickly that it does not allow enough time for it to be auto-extended (or manually extended if you prefer). If you are not using over-provisioning (creating logical volumes in excess of the size of the backing thin-pool), then be prepared to remove snapshots as necessary if the thin-pool begins to near capacity.