gluster-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Gluster-devel] RPM re-structuring


From: Anand Avati
Subject: Re: [Gluster-devel] RPM re-structuring
Date: Sun, 28 Jul 2013 12:19:27 -0700



        Another model can be:

        0. glusterfs-libs.rpm - libglusterfs.so libgfrpc.so libgfxdr.so
        1. glusterfs (depends on glusterfs-libs) - glusterfsd binary,
        glusterfs
        symlink, all common xlators
        2. glusterfs-rdma (depends on glusterfs) - rdma shared library
        3. glusterfs-geo-replication (depends on glusterfs) - geo-rep
        related
        objects
        4. glusterfs-fuse (depends on glusterfs) - fuse xlator,
        mount.glusterfs
        5. glusterfs-server (depends on glusterfs) - server side
        xlators, config
        files
        6. glusterfs-api (depends on glusterfs-libs) - libgfapi.so and
        api.so
        7. glusterfs-resource-agents (depends on glusterfs)
        8. glusterfs-devel (depends on glusterfs-libs) - header files for
        libglusterfs
        9. glusterfs-api-devel (depends on glusterfs-api) - header files
        for gfapi

        This way qemu will only pick up libgfapi.so libglusterfs.so
        libgfrpc.so
        and libgfxdr.so (the bare minimum to "just execute") for the
        binary to
        load at run time. Those who want to store vm images natively on
        gluster
        must also do a 'yum install glusterfs' to make gfapi 'useful'.
        This way
        Fedora qemu users who do not plan to use gluster will not get
        any of the
        xlator cruft.


    I like the idea about users of qemu not having to do with
    non-required glusterfs cruft but with this model we still have
    glusterfsd binary being pulled in for consumers who want libgfapi alone.


How? libgfapi depends only on glusterfs-libs. Whereas glusterfsd is in
glusterfs rpm.

In the scenario when somebody tries to use the gluster client xlators via libgfapi, we would end up installing glusterfsd still. Do we really need that to happen?


Isn't the concern that someone who is not interested in using glusterfs is now forced into removing qemu, libvirt and the whole virt shebang when trying to uninstall glusterfs rpm? glusterfsd coming along when common xlators (client xlators) are installed is probably not a big deal.. That would happen only for users who are interested in glusterfs anyways.

Avati

reply via email to

[Prev in Thread] Current Thread [Next in Thread]