[rescue] The State of SunHELP (new server build, etc)

[Mike Spooner]

Yes, according to https://lwn.net/Articles/665299/, as of Linux 4.4, a write-ahead journal is (or can be) used to effectively close the RAID5/6 write-hole; although I am a little confused as to how the journal itself is protected...




Regards,


Mike




On Sun, Jun 16, 2019 at 8:33 PM +0100, "Andrew Jones" <andrew at jones.ec> wrote:










On 6/14/19 7:51 PM, Bob Darlington wrote:
> zfs raidz specifically eliminates the write-hole issues.  On the smallest
> of my systems I'm running 20 8TB disks raidz3 with a hot spare, times 4 for
> 84 disks in the jbod, and typically 4 of these craptastic seagate jbods to
> a server.  I can't vouch for smaller systems, but I will say my stuff is
> fairly quick.  Here's some of the settings...

Regular ol' Linux mdraid + lvm + xfs also eliminates the write hole, and 
doesn't require any third party out-of-tree drivers with questionable 
licensing.

mdraid + lvm + xfs use the hardware write barrier support built into 
Linux device drivers, because the whole stack is meant to work together.

Speaking of which: does ZFS-on-Linux support the hardware write barriers 
at all?

I know it did on Solaris, but ZFSoL is a different animal.
_______________________________________________
rescue list - http://www.sunhelp.org/mailman/listinfo/rescue