Ed Greshko writes:
On 01/01/18 06:19, Sam Varshavchik wrote:
> Tom Horsley writes:
>
>> On Sun, 31 Dec 2017 23:45:12 +0800
>> Ed Greshko wrote:
>>
>> > Everything went smoothly with no delay
>>
>> I tend to suspect that the rpcbind delay was caused
>> by having NFS exports and the NFS services running
>> (not that any NFS clients were using the service).
>
> I do not use NFS. systemd still took its sweet time closing and reopening
the
> listening socket.
>
The question is then what are you running that requires rpcbind? On my
systems
without an active NFS service it isn't enabled or running.
rpcbind is directly or indirectly required by libvirt. dnf wants to remove
libvirt, if I tell it to remove rpcbind. I'm using KVM to run a bunch of
Window VMs.
That answers the question of why I have rpcbind installed.
Looking at why it is enabled:
[root@monster multi-user.target.wants]# cd /etc/systemd/system/multi-user.target.wants
[root@monster multi-user.target.wants]# ls -al rpcbind.service
lrwxrwxrwx. 1 root root 35 Nov 19 2011 rpcbind.service ->
/lib/systemd/system/rpcbind.service
My rpcbind service seemed to have been enabled in 2011. Its current vendor
preset is "disabled". So, I guess, it wouldn't be enabled today, by default.
Was it enabled by default in 2011? No idea. What does disabling it would
mean, with respect to libvirt, and any guest VMs? No idea.
... Hold on a second. Digging through Wikipedia, Fedora 16 was released a
week before the rpcbind's enabling timestamp on my server.
Grabbed
https://archives.fedoraproject.org/pub/archive/fedora/linux/releases/
16/Fedora/x86_64/os/Packages/rpcbind-0.2.0-15.fc16.x86_64.rpm
postinstall scriptlet (using /bin/sh):
if [ $1 -eq 1 ] ; then
# Initial installation
/bin/systemctl enable rpcbind.service >/dev/null 2>&1 || :
fi
Well, that answers that. Happy New Year, everyone!