On 29/01/18 15:19, NeilBrown wrote: > Hi Ian, > following are three fixes for the autofs patches. > The first is a revised version of a patch were have been > discussing that relates to use_hostname_for_mounts. > Rather than comparing weight and path and hostname etc to see > if two 'struct host' came from the same config-file entry, > I thought it would be best to record and index number and > just compare that. Hopefully that makes it more obvious > what is happening. Right, as I say, this was somewhat broken and what you're recommending is an improvement although I think it doesn't quite fix the possible problems either. It's worth applying when it finally arrives. > > The second fixes that monotonic_elapsed() problem we discussed at > LCA. Customer has confirmed that this fixes the problem. Applied, it's in the queue. > > The third fixes a small Makefile bug Yeah, the config files should be backed up but two installs will cause problems. Most distribution packaging will want to do that step separately so it does look like a sensible thing to do. This is also in my patch queue. I'm going to need to post the patches for RFC and commit them fairly soon and possibly another release, there's a deadlock with the "automount -m" command that needs fixing in 5.1.4. Thanks Neil. > > Thanks, > NeilBrown > > > --- > > NeilBrown (3): > use_hostname_for_mounts shouldn't prevent selection among replicas > Fix monotonic_elapsed. > Makefiles.rules: remove 'samples' from SUBDIRS > > > Makefile.rules | 2 +- > include/replicated.h | 3 ++- > lib/rpc_subs.c | 4 ++-- > modules/mount_nfs.c | 2 +- > modules/replicated.c | 35 ++++++++++++++++++++--------------- > 5 files changed, 26 insertions(+), 20 deletions(-) > > -- > Signature > -- To unsubscribe from this list: send the line "unsubscribe autofs" in
