> Can you please provide me the details as to how did
> you manage to compile QFS 5.0 on Opensolaris. I
> downloaded the sam-qfs src code from ssh://anon at hg
> dot opensolaris dot org/hg/samqfs/samqfs, but when I
> try to compile, I get
> 
> make[2]: Entering directory
> `/local/chak/sam-qfs/samqfs/src/robots/vendor_supplied
> '
> [ -d ../../../lib/obj/SunOS_5.11_i386_DEBUG ] ||
> mkdir -p ../../../lib/obj/SunOS_5.11_i386_DEBUG
> make -C stk 
> make: Entering an unknown directory
> make: *** stk: No such file or directory.  Stop.
> make: Leaving an unknown directory
> make[2]: *** [stk] Error 2
> make[2]: Leaving directory
> `/local/chak/sam-qfs/samqfs/src/robots/vendor_supplied
> '
> make[1]: *** [robots/vendor_supplied] Error 2
> make[1]: Leaving directory
> `/local/chak/sam-qfs/samqfs/src'
> make: *** [src] Error 2

To work around that problem, I simply removed the "robots" entry in 
"samqfs/src/Makefile" (or was that GNUmakefile?)

Also, there are references to SANergy under the samfs/fs section.  Those need 
to be removed from the C files.  IIRC there are three entire methods that need 
to be deleted or commented out.

And when you do get past all that, you'll have to go through all the 
subdirectories under samqfs/src/lib (IIRC) and make about half the libraries by 
hand by running gmake in that directory.  There were also dependency problems 
that I found I could only resolve by running gmake several times in a row.

I finally got to the point where I created a script similare to this:

for makefile in `find . -name GNUmakefile`; do
    gmake -C `dirname $makefile`
done

I'd run that three or four times until everything built properly.

It didn't take me long to get QFS to compile, but I've got quite a bit of 
experience with large C/C++ applications that involve kernel modules.  I also 
pretty much know SAM/QFS about as well as someone who's not a Sun developer 
employed directly on SAM/QFS can.  In fact, supporting SAM/QFS installations 
has been at least a part of my day job for probably 6 or 7 years now.  That 
experience was important in taking the QFS binaries that the make process 
creates and piecing together a working install.  The basic QFS 5.0 installation 
that I got working is pretty similar to a QFS 4.6 installation.  I don't recall 
any significant changes in file locations in the /opt/SUNWsamfs, 
/usr/lib/fs/samfs, /etc/opt/SUNWsamfs, and /var/opt/SUNWsamfs directories.

The most vexing issue I had was one of the kernel modules doesn't have a 
supplied .conf file.  And without that, it won't attach.  Without any error 
messages as to why.  IIRC, you can just copy the .conf file over from QFS 4.6.

Finally, all I know that I have working is the basic QFS file system 
functionality.  I can guarantee you that SAMFS functionality isn't going to 
work, nor is any SANergy integration, although I suspect that if you could get 
the proper libraries from a SANergy installation you might get it working,  But 
you'd be limited to SPARC platforms.

Whether or not shared QFS works with the released code is still unanswered.  
FWIW, the one cursory examination I did of the QFS 5.0 implementation of shared 
file systems seems to indicate significant changes from QFS 4.X.
-- 
This message posted from opensolaris.org
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss

Reply via email to