Not necessarily all, but at least after most times that i run
sudo zypper dup --no-allow-vendor-change
, there’s a message at the end advising that some programs are running that are now using deleted files, & so i “might want to” restart them. Is that really just an indirect way of telling me that i should at least log out/in, or even reboot, rather than continuing just to use TW without interruption? Or, to put it another way, do i risk “bad things happening” if i did simply keep working without either doing a logout or reboot?
On 07/09/2017 09:56 AM, GooeyGirl wrote:
>
> Not necessarily all, but at least after most times that i run
> Code:
> --------------------
> sudo zypper dup --no-allow-vendor-change
> --------------------
> , there’s a message at the end advising that some programs are running
> that are now using deleted files, & so i “-might want to-” restart them.
> Is that really just an indirect way of telling me that i should at least
> log out/in, or even reboot, rather than continuing just to use TW
> without interruption? Or, to put it another way, do i risk “bad things
> happening” if i did simply keep working without either doing a logout or
> reboot?
>
>
My rule of thumb for this is:
If a process is owned by root, reboot
if not log out and in.
–
Ken
linux since 1994
S.u.S.E./openSUSE since 1996
Hi
If all the processes are your user (zypper ps -s) then a login/logout should suffice, depends on the others and services running, especially if it’s a kernel update (not a initrd rebuild) worth a reboot IMHO.
If you follow the factory mailing list you can see what changes are for, for example;
Add some verbosity so you can see what is changing (you can also see from what repo as well to highlight any funnies)
zypper -vvv dup --no-allow-vendor-change
I also don’t use sudo… just su - to switch to root user.
There was one small update around 1 week ago where I did not reboot.
If there’s a kernel update – reboot;
if there’s a “systemd” update – reboot;
if there’s an update to core system functions, such as “udev” – reboot.
if there’s an update to your desktop environment – at least logout and login (or reboot).
Most Tumbleweed update change enough that reboot seems about right. If this is an inconvenient time to reboot, then maybe I should postpone that update until a more convenient time.
On Sun, 09 Jul 2017 14:26:01 +0000, malcolmlewis wrote:
> Hi If all the processes are your user (zypper ps -s) then a login/logout
> should suffice, depends on the others and services running, especially
> if it’s a kernel update (not a initrd rebuild) worth a reboot IMHO.
Does TW not use the kernel update mechanism that doesn’t require a reboot?
For me (on Leap), I use zypper ps -s to check the processes and restart
them manually. The one thing that I will occasionally trip over on one
of my systems is if I forget to suspend VBox VMs and there’s a VBox
update, the VBoxManage command won’t see the running VMs any more, and
that’ll force me to shut them down manually.
Another gotcha is that if you restart dbus, it’ll kill the X session, and
you may need to restart systemd-logind, even if it doesn’t show as
needing an update (otherwise login times get quite long).
Lastly, systemctl daemon-reexec is useful when systemd is updated.
When I was young, was so much younger than todayay ♫♪
I used to consider a reboot of a server as not done, ‘uptime’ was directly related to stability etc. Not strange for someone coming from a AT&T Unix System V, where the NCR machine I managed to ~20 minutes to shutdown, ~45 minutes to boot ( memory tests etc ). These days my cloud server is a self-managed VM in a data centre and a reboot takes less than 10 seconds. So, running Tumbleweed on it is not really an issue in my case. And that’s what I do after ( almost ) any TW update. Simply wait until late in the evening, update all machines usiing salt, rebooting 2 machines through salt. Laptop shuts down when I go to bed.
As per today the ‘–no-allow-vendor-change’ is a default zypper option for TW, so a plain ‘zypper dup’ is enough. I already had that configured manually
Thank you all for your replies, & sorry for my delayed reply – pseudo-real-life stepped in for a couple of days.
What you’ve all said is pretty much what i’d already worked out to do myself & have been doing], but i decided to ask anyway just in case i was performing reboots when not actually necessary.
As i wrote elsewhere in these fora, oS TW is my first honest to goodness rolling release exposure, & i’ve been feeling conflicted about how to manage myself/TW given that i generally dislike frequent reboots due to the workflow disruption, but conversely i’m not going to knowingly endanger my system. Left to my own devices, my pre-TW usage-paradigm was to leave my various open & still needed pgms & in-progress docs distributed around my various VDs & Activities, then before bed at night Suspend not Hibernate; i don’t/can’t use that due to encrypted home & swap] the pc. Come morning i Resume, & can get “straight back into it”; i don’t have to faff about reopening all my stuff, reallocating said stuff to the applicable VD & Activity, finding where i was up to, etc. Obviously reboots totally conflict with this desired workflow. Given the TW updates occur either daily, every other day, or every other other day, i’ve found it very disruptive. Hence, i’ve decided to let myself only manually check for updates now on a weekly not daily basis; this feels a least-worst compromise between still “getting my new TW toys & goodies” relatively fast vs actually getting on with my stuff efficiently.
Oh, btw, i meant to ask this weeks ago but forgot… by default in the Plasma System Tray there’s a Software Updates icon/widget/thingie. Most of the time it’s useless as it insists that i’m offline when [of course] i’m not, but occasionally it wakes up & realises there is an internet, & then tells me if there’s updates. I used to do my updates this way until i read the sometimes-heated historical posts elsewhere about what the “right” way is to update TW, which is when i adopted the cli method per the topic subject. What is the right thing to do with this Plasma widget - use it, or remove it so as to not be tempted by it? [some of the posts i’d read implied that using this widget performed an [i]up not a dup, & was unsafe for TW]? The arguments & counter-arguments quite confused me.
Oh yes, thanks – indeed i do know how to hide it, or to actually instead remove it… but i was trying to understand if either tactic was advisable/necessary. Ie, if it does do a up not dup, then presumably [being TW not Leap] that would be ample reason to get rid of it…?
(then no temptation to use it)