[Linux-cluster] NFS on GFS architectural issues / problems

[Robert Peterson]

Riaan van Niekerk wrote:
> hi Bob and others
>
> I found on the Red Hat 108 Developer Portal the following GFS1/GFS2 
> design document which details amongst others, some of the issues with 
> NFS on GFS:
> https://rpeterso.108.redhat.com/servlets/ProjectDocumentView?documentID=99 
>
>
> (I see it was sent to this list over a year ago, but I never found it 
> while searching through the archives. it has a lot of good information 
> in it)
>
> It has a disclaimer: Some of the comments
> are no longer applicable due to design changes
>
> My question to you or anyone who is familiar with NFS on GFS, or GFS 
> in general, which of the following are still valid issues for the 
> current (6.1u4) version of GFS. If all or most of them still apply, I 
> can use this as motivation for my customer to strongly consider going 
> off NFS on GFS. Removing the NFS from our GFS cluster has been on the 
> cards for quite a while, but has not gained momentum due to lack of 
> information on the performance gains of such a move (very difficult to 
> gage) or the architectural problems/limitations of NFS on GFS (for 
> which the following extract is spot-on).
>
> Note - can you consider adding a link to this document from your FAQ?
Hi Riaan,

The document you mentioned was written by Ken Preslan more than a year ago.
It has some good architectural information regarding GFS and GFS2, but 
the problem is, there
have been a lot of changes to GFS2 and a lot of work has been done on 
NFS since that time,
so a lot of it no longer applies.

One day I was playing with 108 and decided to upload the document to my 
108 page because I thought it
was "a good find" and there was a need for  GFS architectural 
information on the Internet.
Afterward, I was discussing the article with some of the developers and 
they all agreed
that the article shouldn't be posted because it contained too much 
misinformation due to recent changes
made to all areas of the code.  The problem is, I had already posted the 
article and I couldn't figure
out how to get 108 to delete it.  (Today I figured out how to delete it, 
and did so, and I apologize if
anyone was misled by what it says.  I'm going to file a usability 
bugzilla against 108 though.)

What I really need to do is write a white paper about GFS and its 
internals and its structures,
rather than spending the time required to sift through Ken's article and 
separate fact from "no longer
applicable".  And the link to that document will certainly be added to 
the FAQ.

There are some known issues with NFS failover, but it works great unless 
you're intentionally trying to break it
by doing some nasty tricks such as those documented in bugzilla 178057.  
If you read the first few comments
of the bugzilla, you'll see that I tried very hard at first to break it 
and couldn't.
Wendy Cheng has been spearheading the effort to improve NFS failover and 
I applaud her efforts.

Regards,

Bob Peterson
Red Hat Cluster Suite