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

PATCH rhel6-branch] (take II) network: using more than one NICs (iSCSI) (#638131, #634016, #668395)



This is take 2 of a patchset fixing 3 newtork configuration bugs in loader:

* https://bugzilla.redhat.com/show_bug.cgi?id=638131
  "Cannot perform network kickstart installation to iSCSI target accessed via different NIC"
  Patches 1/11 - 8/11.

  With the patches, anaconda should be able to fetch repositories via one
  NIC while accessing iSCSI via different NIC using kickstart configuration.
  All connections are activated in loader.

  These things were needed:

  1) Activate more than one device in loader using ks -> changes
     in loader and --activate option is added to kickstart network
     command so that former behaviour (only configuring additional
     devices from ks) is preserved.
  2) Support for network --bootproto ibft is added into loader.
     (in scope of this bug still writing out values which we read
     from ibft in anaconda into ifcfg files, see 4)).
     In stage 2 ks - where only device configuration (not activation)
     can happen in case the device was not activated with --activate
     option in loader - the option would be supported automatically
     when we start to write out BOOTPROTO=ibft in loader (see 4) below)
  3) way to prevent device activated using ks to grab default route ->
     --nodefroute option is added (setting DEFROUTE=no in ifcfg file)

  Kickstart part was acked in take 1 of the patchset, I cloned pykickstart
  to trace it in the component (#668417).

  Just a note for patch reader - in 4/11 there is a hunk @@ -2046,24 +2062,10 @@
  where only unreachable code is removed.


* https://bugzilla.redhat.com/show_bug.cgi?id=634016
  Patch 9/11.

  4) Use NM's BOOTPROTO=ibft and dracut's ip=ibft instead of passing
  values obtained by anaconda to them. This requires one NM fix:
  https://bugzilla.redhat.com/show_bug.cgi?id=665027 and assumes that:
  - BOOTPROTO=ibft works fine in NM:
    https://bugzilla.redhat.com/show_bug.cgi?id=479824. It did in
    my tests in installer environment, except for the issue mentioned
    in 4). I was not able to test post install behavior as I couldn't
    get over dracut.
  - ip=ibft works in dracut:
    https://bugzilla.redhat.com/show_bug.cgi?id=640979
    My tests with rhel6 nightly of Dec 20 didn't work. I'll investigate
    why.


* https://bugzilla.redhat.com/show_bug.cgi?id=668395
  "Kickstart network command doesn't reconfigure active connection (unlike in rhel 5)"
  Patches 10/11, 11/11.

  The behaviour of rhel 5, rhel 6.0, and the desired one for 6.1 is
  described here: https://fedoraproject.org/wiki/Anaconda/Network#kickstart.

  For the 1st and the 3rd bug documentation should be updated and they
  probably require release notes.



Radek


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