On 07/31/2018 10:23 AM, Francis.Montagnac(a)inria.fr wrote:
Hi
On Tue, 31 Jul 2018 09:35:27 -0700 Rick Stevens wrote:
> If you don't have the automount stuff, I think the most reliable thing
> is to give your client a fixed IP/netmask/gateway combination. ...
This may not be sufficient: I had a client with fixed IP configuration
that fails to mount an NFS server at boot.
The turnaround that works for me is to ovverride
NetworkManager-wait-online.service to disable the use of nm-online and
to wait instead for the (static) routes/gateways to become
defined/reachable.
See the thread:
NetworkManager-wait-online is still utterly, and completely, broken
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.o...
and my post:
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.o...
Yes, I realize this. This is one of the issues with systemd in general,
it parallelizes stuff and causes these race conditions. The entire
network startup process should be single threaded (IMHO) to allow the
appropriate sequence of events to occur before stating that the network
is either up or down.
I reiterate my previous statements: systemd is a solution in search of a
problem. Parallelizing boot processes to speed things up is sorta silly.
Typically, booting a system is a sequence of operations that must be
performed in a specific order. A huge part of the complexity of systemd
(oneshots, forks, wait-for-this-or-that) is to try to get around its
attempts to parallelize these processes.
If you have to reboot so often that boot speed is an issue, your system
has other issues.
----------------------------------------------------------------------
- Rick Stevens, Systems Engineer, AllDigital ricks(a)alldigital.com -
- AIM/Skype: therps2 ICQ: 226437340 Yahoo: origrps2 -
- -
- If it's stupid and it works...it ain't stupid! -
----------------------------------------------------------------------