update-manager --no-focus-on-map ??

[Liam Proven]

On 1 January 2016 at 19:00, Ralf Mardorf <silver.bullet at zoho.com> wrote:
> rebooting after each upgrade at least makes the hard disk drives spin
> down and up, in other words it causes unloading and loading the heads
> and this will shorten the lifetime of hard disk drives

You are *really* stretching the point there. I do not believe that you
could demonstrate this as a real issue with any amount of
instrumentation or the finest possible precision of measurement.

No. Rejected.

People often get obsessed with theoretical performance or lifetime
measurements that are not true in real-world use.

So, for example, 20y ago, I was running a testing lab for one of the
main UK computer magazines. I also built that magazine's 32-bit
benchmark suite, porting it from their original 16-bit benchmark
suite.

At that time, Windows 95 came out.

Now, remember, Windows 95 *A* -- the original release -- did not
include FAT32. It only supported FAT16. FAT32 appeared with Windows
95B, known as OSR2: OEM Service Release 2. A big feature for a service
release!

At that time, 1.2GB was a normal hard disk size. Formatting a >1GB
drive with FAT16 means 32kB clusters, which are very very inefficient.

I wrote in the magazine that if you reformatted with 1 x 0.99GB
partition, you would actually have more overall disk space on drive C
afterwards. If you then used the leftover space for a small 200MB
partition for drive D:, and put the Windows swap file on that drive D,
you'd have nearly half a gig more free space -- for nothing. Totally
for free.

Several readers wrote in, and several PC manufacturers, complained.
Putting the swap file on the extreme edge part of the disk like that,
which had measurably slower access times and read times, would reduce
PC performance.

When it was readers, I replied in the magazine and said that the
effect would be too small to measure.

But when PC makers complained when I marked their PCs down in reviews
for not doing it, I had to prove it.

So I invited some into the labs and did a live demo. The same PCs were
benchmarked with and without a separate swap drive.

Of course, it made no difference at all, not even to the 2nd decimal
place of 1 percentage point.

But people get hung up on differences that you cannot measure.

Your shortening of HD life time by too many reboots is in this area.
It is a nonsensical objection; it would have no measurable effect.

That is ignoring the increase in prevalence of SSDs.

(SSDs are a different matter; switching to a non-journalling
filesystem and disabling access time recording /does/ measurably
increase the performance and the life of Flash media.)

> while rebooting
> without a good reason, doesn't gain something useful.

Yes it does if you have a rogue update process taking memory and RAM.

That was the OP's question, remember.

> I suspect that
> assumed an upgarde requires a reboot, there will be a warning a user
> won't miss, assumed the user doesn't use "assume yes" options.

It will ask for one, yes. But rebooting if it is not inconvenient does
no harm at all.

> Not replacing the ";" with "&&" between update and dist-upgrade is ok
> too, as well partial upgrades that require missing dependency upgrades
> won't happen, assumed an update should fail. Anyway, it's preferable
> that users get used to ensure, that all index files were re-synced.

Agreed, but if it's easier to type a shorter simpler command and not
explain this, then that is what I myself do and recomment.

> Good
> habits one day might be helpful, if for what reason ever, a user
> should migrate to another distro or unix-alike operating system.

It would be nice, yes. But remember that Ubuntu is not just another
Linux. It doesn't even have "Linux" in its name. It's "for human
beings", in other words, *non* geeks, *non* techies.

Trying to instil good tech habits in non-technical people is hard work
for no reward.

> Avoiding "assume yes" options is best practices what ever distro,
> operating system, application is used. As a matter of principle, spending
> a few seconds to read a few words, even if a users shouldn't understand
> everything, doesn't harm, so users e.g. don't miss warnings. Btw. how does
> "assume yes" act, if a package starts a script that requires interaction
> to set up configurations? For good reasons, the Ubuntu default is _not_ to
> assume yes.

Again, it would be nice, but easy is preferable to theoretically
correct but harder.

Sad but true.

Surely you know the classic essay?

https://en.wikipedia.org/wiki/Worse_is_better

> To use a new kernel I prefer to reboot, anyway, there is a FLOSS method
> available to switch to another kernel without rebooting.
> https://en.wikipedia.org/wiki/Kexec
> http://manpages.ubuntu.com/manpages/wily/man8/kexec.8.html
> http://packages.ubuntu.com/wily/kexec-tools

True, but it's not widely-used yet.

Kexec is technically a reboot, anyway. It doesn't rerun the firmware
POST but it completely reloads your OS. It's still a reboot, just a
faster one.

I was referring to things like Ksplice, Kpatch and Kgraft.


-- 
Liam Proven • Profile: http://lproven.livejournal.com/profile
Email: lproven at cix.co.uk • GMail/G+/Twitter/Flickr/Facebook: lproven
MSN: lproven at hotmail.com • Skype/AIM/Yahoo/LinkedIn: liamproven
Cell/Mobiles: +44 7939-087884 (UK) • +420 702 829 053 (ČR)