On Fri, Feb 05, 2010 at 11:55:12AM -0800, Bob Friesenhahn wrote: > On Fri, 5 Feb 2010, Miles Nordin wrote: > > > > ls> r...@nexenta:/volumes# hdadm write_cache off c3t5 > > > > ls> c3t5 write_cache> disabled > > > > You might want to repeat his test with X25-E. If the X25-E is also > > dropping cache flush commands (it might!), you may be, compared to > > disabling the ZIL, slowing down your pool for no reason, and making it > > more fragile as well since an exported pool with a dead ZIL cannot be > > imported. > > Others have tested the X25-E and found that with its cache enabled, it > does drop flushed writes, but is clearly not such a gaping chasm as > the X25-M. Some time has passed so there is the possibility that > X25-E firmware has (or will) improve. If Sun offers an X25-E based > device for use as an slog, you can be sure that its has been qualified > for this purpose, and may contain modified firmware. > > The 'E' stands for "Extreme" and not "Enterprise" as some tend to > believe. >
I missed out on this thread. How would these dropped flushed writes manifest themselves? Something in the logs, or just worsened performance? Ray _______________________________________________ zfs-discuss mailing list zfs-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
