Re: ACL

On Aug 26, 2006, at 12:41 PM, André Aulich wrote:

Hi, David,

On 25.08.2006, at 01:14, David Thompson wrote:

Yes, but obviously the issue here is that if an MDC fails, there should be fail over on the volume. The fact that an MDC doesn't fail over to a replica DC is a big issue.

In my setup MDC failover worked, yet, depending on the test scenario it took almost no time (if you follow Apple's recommendations: http://docs.info.apple.com/article.html?artnum=301787) or up to several minutes (with an Xsan volume mounted on the failing MDC).

I guess with Xsan 1.4 Apple enforces its former recommendations – which makes sense for a high availability system like Xsan.
I was just surprised, because I haven't read anything about it...

However I just ran the same test as you on my own MDC (unmount the volume on 1 MDC which had it mounted, then shutdown the server) I then stopped the volume and it stopped properly. I then started the volume and it started and I was able to mount it without issue on my 2nd MDC. I then brought back up the powered down MDC and everything is working as advertised.

Perhaps you have a bum install of Xsan. I know I had issues (as did many other people) with Xsan 1.3

No, all four systems I run with Xsan 1.4 work great. When I run the same test like yours I see no problems.
It only happens when I restart both MDCs.

Well are you talking about rebooting both MDC's at the same time? If so then certainly your SAN would not be available until the first MDC to reboot and make the SAN volume(s) available is up and running. Is this what you are talking about? Shutting both MDC's down and only leaving the san clients on?

References:
	>ACL (From: Mark Nassy <email@hidden>)
	>Re: ACL (From: Gustavo Beathyate <email@hidden>)
	>Re: ACL (From: Mark Nassy <email@hidden>)
	>Re: ACL (From: André Aulich <email@hidden>)
	>Re: ACL (From: André Aulich <email@hidden>)
	>Re: ACL (From: André Aulich <email@hidden>)