Hi all,
It's finally time for a release, autofs-5.1.4.
Unfortunately there were a couple of annoying regressions in 5.1.3.
I believe they have been resolved in this release.
Also, the startup retry on the master map not available logic has
been improved so that should function properly now and help with
reported problems of startup failures. The delay is configurable
in case the default is not sufficient.
A new amd mount type has been added. The amd mount type "program"
has been added which allows a user provided program to be given
in a map entry that will be used for performing the mount and
optionally another program to be used at umount.
Additionally there are a number of bug fixes and other minor
improvements.
autofs
======
The package can be found at:
https://www.kernel.org/pub/linux/daemons/autofs/v5/
It is autofs-5.1.4.tar.[gz|xz]
No source rpm is there as it can be produced by using:
rpmbuild -ts autofs-5.1.4.tar.gz
and the binary rpm by using:
rpmbuild -tb autofs-5.1.4.tar.gz
See the README.amd-maps file for information about using amd format
maps.
Here are the entries from the CHANGELOG which outline the updates:
19/12/2017 autofs-5.1.4
- fix spec file url.
- fix unset tsd group name handling.
- Add -c option when calling /bin/umount - if supported.
- remove some redundant rpc library code.
- add port parameter to rpc_ping().
- dont probe NFSv2 by default.
- add version parameter to rpc_ping().
- fix typo in autofs config file comments.
- fix typos in autofs man pages.
- use pkg-config to search for libtirpc to fix cross-compilation.
- fix incorrect status return in get_nfs_info().
- fix a couple of compiler warnings.
- set systemd KillMode to process.
- fix mount.nfs blocks on first mount.
- fix some man page problems.
- add some more debug logging to get_nfs_info().
- add some more debug logging to get_supported_ver_and_cost().
- fix ipv6 proto option handling.
- also check flag file exe name.
- fix possible map instance memory leak.
- check map instances for staleness on map update.
- allow dot in OPTIONSTR value lexer pattern.
- fix autofs_use_lofs description.
- fix amd parser error buffer size.
- make spawn_bind_mount() use mount_wait as well.
- document ghost option in auto.master man page.
- only take master map mutex for master map update.
- revert fix argc off by one in mount_autofs.c.
- reset master map list on startup retry.
- fix nisplus lookup init not configured check.
- make open_lookup() error handling more consistent.
- be silent about sss library not found.
- be silent about nis domain not set.
- make map source reference message debug only.
- improve description of mount_nfs_default_protocol.
- the port option should not behave like nobind option.
- handle additional nfs versions in mount_nfs.c.
- fix symlink option passthrough in mount_nfs.c.
- fix ordering of seteuid/setegid in do_spawn().
- update configure to check for pipe2(2).
- fix open calls not using open_xxxx() calls.
- move open_xxxx() functions to spawn.c.
- serialize calls to open_xxxx() functions.
- improve debug logging of lookup key.
- fix typo in amd_parse.c.
- add missing MODPREFIX to logging in amd parser.
- fix symlink false negative in umount_multi().
- remove expand_selectors() on amd parser entry.
- fix amd defaults map entry handling.
- refactor amd_parse.c.
- fix amd parser double quote handling.
- fix expandamdent() quote handling.
- fix possible memory leak during amd parse.
- remove path restriction of amd external mount.
- add function umount_amd_ext_mount().
- add function ext_mount_inuse().
- add function construct_argv().
- add amd mount type program mount support.
- fix memory leak in umount_amd_ext_mount().
- fix strerror_r() parameter declaration in do program_mount().
- fix incorrect check in validate_program_options().
- fix log message in spawn_umount().
- workaround getaddrinfo(3) ai_canonname bug
Ian
On 12/19/2017 02:29 AM, Ian Kent wrote:
> A new amd mount type has been added. The amd mount type "program"
> has been added which allows a user provided program to be given
> in a map entry that will be used for performing the mount and
> optionally another program to be used at umount.
>
What's a use case for this?
On 19/12/17 17:45, Patrick Goetz wrote:
> On 12/19/2017 02:29 AM, Ian Kent wrote:
>> A new amd mount type has been added. The amd mount type "program"
>> has been added which allows a user provided program to be given
>> in a map entry that will be used for performing the mount and
>> optionally another program to be used at umount.
>>
>
> What's a use case for this?
It was requested and it is one of the standard amd mount types
that hadn't been implemented. If nothing else it makes a slightly
more complete amd map format implementation.
I've seen a case where someone used a script to customize a mount
command to suit different environments, where the mount command
needed to be slightly different for different platforms. Probably
not so useful to autofs but am-utils can run on a number of different
OSes so the amd maps used by autofs (ideally unchanged from other arches)
need to be able to cope with this.
There's also the case where a mount needs to use a custom command to
perform the mount (and possibly the umount at expire).
Long ago this was requested for Sun format maps but I'm still not
in a position to implement that (and it isn't part of the defined
Sun format map syntax either).
Ian