Given persistent l2arc there are definite performance reasons to move the cache with the pool. From a performance standpoint loss of the cache device is not harmless. Not tragic but also not harmless.
Sent from my iPhone > On Sep 11, 2015, at 7:20 AM, Andriy Gapon <andriy.ga...@clusterhq.com> wrote: > >> On 11/09/2015 16:47, George Wilson wrote: >> The ideas was that only one pool could own it so that failover would be >> possible and the cache device would follow that pool. It also would be >> required for persistent l2arc. >> >> However, it should be possible to create an independent l2arc which never >> fails over but can be added to multiple pools. It would not be a trivial >> implementation but seems doable. > > Well, given that the cache device has just a cache of data it can easily > follow > its pool by simply discarding the cached data. > Loss of a cache device is harmless even now. > > -- > Andriy Gapon > _______________________________________________ > developer mailing list > developer@open-zfs.org > http://lists.open-zfs.org/mailman/listinfo/developer _______________________________________________ developer mailing list developer@open-zfs.org http://lists.open-zfs.org/mailman/listinfo/developer