[NTLUG:Discuss] Disaster recovery

[Bryan J. Smith]

Chris Cox <cjcox at acm.org> wrote:
> Even the difference between a 32-bit Xeon and a 32-bit Xeon
> with HT can be an issue...

Really?  I'm intriqued.  Do you have an example?

> Certainly a problem if the other processor is EM64T or AMD
> (lookout!).

???  I'm not trying to read too much into that, but ... ?

> Chipsets can mean the difference between a working DMA and
> a non-working DMA... can mean the difference between seeing
> a harddrive and not seeing one.

I avoid on-board ATA at all costs.  Not just for DMA issues
in Linux, but because of volume issues.  Heck, even NT5+
(2000+) seems to set 3 registry settings that are
model-specific.

For ATA, I like to stick with 3Ware because older volumes are
always readable on newer cards/firmware.  Even for JBOD, they
really get rid of the typical issues.

> ACLs are still very non-standard across *ix...

The IEEE-X/Open "Austin Group" that created the Single UNIX
Specification (SUS) v3 and newer (2001+) IEEE POSIX standards
layed down the gauntlet on ACLs.

In Linux, ACLs were standardized in kernel 2.5.3 and both
Ext3 and XFS adopted the standard.  Most of the current ACL
code in 2.6, and backported to 2.4, is based on the merger of
both XFS and Ext3-EA code.

> but if we're talking just Linux.. then you just have to
make
> sure to use a backup format that can handle ACLs.  I know 
> that s-tar handles them (though it had some serious bugs
> until very recently)...

star is adequate for Ext3, but as you said, it has bugs for
Ext3 and a couple other platform/filesystems.

I prefer "xfsdump" because XFS can store upto 64KiB of
meta-data per inode directly.  XFS' structure, including its
EA/ACL stores, remain unchanged since the mid-'90s, and its
Linux port is direct from Irix in this regard.

> not sure if GNU tar has added adequate support or not.

GNU Tar has been largely non-POSIX compliant due to
limitations in the old POSIX (circa 1989) standard, but newer
1.14+ releases are trying to address capabilities.

> My soln... kill the extended ACL support.. it's still 
> baking.

My solution since early 2001:
  Adopt XFS for data volumes.  ;->


-- 
Bryan J. Smith                | Sent from Yahoo Mail
mailto:b.j.smith at ieee.org     |  (please excuse any
http://thebs413.blogspot.com/ |   missing headers)