Linda Walsh said the following on 04/12/2013 09:54 PM:
Anton Aylward wrote:
Linda Walsh said the following on 04/11/2013 09:59 PM:
Putting all of usr on /rootfs nearly does the same thing, as the reason for a small root was to have less on it needing updating.
You are so missing the point.
Yes, I am. Could you tell me the point?
Because you can still have separate / and /usr without the problems you raise. I'm really baffled as to why you and others don't drill down and read the back-story and admin notes and guides that surround the systemd project.
To me it creates a more unstable system, more prone to failure and more difficult to restore to running condition.
How so? You say that but don't say how that is so.
This means more downtime and lower reliability overall. That's my point. So why do we want that?
The only way I can see that you'd zap a / and not /usr of they were separate is - BTDT - finger trouble. Else your whole disk dies. Recovery modes reflect failure modes. Right now, I'm running 12.3 on a old (so old the surface ought to be flaking off the platters) 20G drive. Actually a couple of them in old 800 MHz machines out of the Closet of Anxieties. On one I have, as I said, the whole system that isn't from NFS, which means that the basic / and /usr are there. Crappy SiS video. With the latest kernel its still acceptably fast. On the other everything except /boot is LVM and there is separate / and /usr. My point is that I've seen and am using both ways. The difference between thee and mee is that I'm running pretty much "out of the box". My problems, which I've described here in the past, are (a) from the hardware, which is old and crappy, and (b) from idiosyncrasies of the setup - that is they are my doing and not the shortcomings of the system, but problems because I went ahead and did stuff without thinking how it would impact the system. But look: I can put the whole system except for /home and all the custom stuff that might be under /srv for the web services on a 20G drive. Lets see, just a mo while I copy back /usr/share ... and do a bit with the basic /home $ df -h Filesystem Size Used Avail Use% Mounted on rootfs 18G 8.6G 7.8G 53% / devtmpfs 1.2G 36K 1.2G 1% /dev tmpfs 1.3G 12K 1.3G 1% /dev/shm tmpfs 1.3G 440K 1.3G 1% /run tmpfs 1.3G 0 1.3G 0% /sys/fs/cgroup tmpfs 1.3G 440K 1.3G 1% /var/lock tmpfs 1.3G 440K 1.3G 1% /var/run tmpfs 1.3G 0 1.3G 0% /media Of course YMMV, but that's a workstation with gimp, libreoffice and all the other stuff you'd expect on a workstation. Lots of writing tools, lots of Java based tools like UML modellers and XMIND and Freeplane in /usr/local. And its still less than 10G. That's with the unified / and /usr That's about twice the size of the UNIX V7 I ran on a PDP-11/45 back in 1982 and that didn't have networking, X11 or Java. Of course the physical disk is a fraction the size of those old "flying saucer" RK05 packs. So you're running big disks and RAID, but its still beside the point. Your argument is that you want a / that is small; I resume you mean easy to back up and restore. Who whoopee Dee! Mine fits on a USB stick That's not just my / but my / and /usr. And not even a stick I had to pay money for but one of the free give-aways I got at a trade show that the vendor puts the product docco on. What? Wait a mo. Here's the SD card from my camera, 32G, or the tiny one from my phone. Here's the USB card reader. No, to be honest, at one point that 20G disk was nearly full, even without /usr/share on it, because I was running 'snapshot' to take images from whenever I ran zypper or altered the config. Snapshot is a great proof against finger trouble (see above) but not a lot of use against your drive dying (see above). Its actually pretty smart about only appearing to eat disk space :-) So, I think you *ARE* missing the point. I don't see any more instability with my 12.3 systems and systemd, the one running separate / and /usr or the unified one, that an do with my 11.4 laptop. When I "zypper up" it updates what it updates and which FS or mounted FS doesn't bother it. The system core and 'essential' applications & libraries are pretty small so backup isn't an issue if you're worried about a disk failure. The Snapshot system is recommended if you fear finger trouble. But what it comes down to, Linda, is that if you've hacked your system about as much as you keep telling us, directly or by implication, then what you say is wrong with the 'out of the box' openSuse (or Fedora or Mageia) that the rest of us are running is pretty much irrelevant because its so far removed from what you're experiencing with your system, and the problems you go on about, which I'm not saying aren't factual for you, arise from the difference between your much modified system and the more vanilla system the rest of us live with. That's the point I think you're missing. All this stuff that your saying is so dreadful seems to work fine for me, even on my crappy hardware from the Closet of Anxieties. Have I had problems? Well yes, but I find that they have, as I said been due to either crappy hardware[1] or my own lack of understanding. Reading and research and thought will cure the former; repeated visits to the Closet of Anxieties or expenditure of hard earned lucre (or bumming someone else's equipment) can address the former. But that's me. The problems some other people face can't be fixed, not even by going back to older distributions. They aren't technical problems. [1] Here's this DVD drive that will read and write just fine but you can't use if for booting. I _think_ its a cable problem. -- "Excellence is not a destination; it is a continuous journey that never ends." -- Brian Tracy -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org