The modules packages produced by module-assistant
do call depmod -a as part of their postinst scripts.

Yeah, I was really saying that the lustre tar source doesn't call
depmod after you run `make install` from an lfs perspective.  Which is
something I've seen done on just about ever other externally built
kernel module package I've ever seen...

In both the client and server cases, the recommended task is
to install lustre-source.
For the client, install lustre-source and use 'module-assistant'. This
should generate the modules deb and install it for you.
In the server case, Unpack a Linux kernel such as 2.6.18, run
make config in it to prepare it, and build using
linux-patch-lustre deb and the Debian Kernel packager,

e.g. in /usr/src/linux-2.6.18, run
# make-kpkg --added_patches lustre -uc -us buildpackage

to build the lustre-patched kernel.

So could I also do these commands on the source from apt-get source
linux-image-2.6.18-4-686? so like

apt-get source linux-image-2.6.18-4-686
make-kpkg --added_patches lustre -uc -us buildpackage
dpkg-buildpackage
<wait overnight for every kernel ever used to build>


Then run this kernel, and build modules for it as in the client case.

This procedure allows you to build optimised server kernels (e.g.
as clean as possible, but with perhaps special drivers for your
hardware), and simple modules installed on top of an ordinary
kernel for the client, which can be useful.

I will put this into a README in the lustre packages for clarity.

Thanks... I keep forgetting that there's those readme's in the debian directory.

Thanks,
- David Brown

_______________________________________________
Lustre-discuss mailing list
[email protected]
https://mail.clusterfs.com/mailman/listinfo/lustre-discuss

Reply via email to