Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > alt.folklore.computers > #213787 > unrolled thread

Where did Microsoft go wrong?

Back to article view | Back to alt.folklore.computers

Contents

  Where did Microsoft go wrong? gareth evans <headstone255@yahoo.com> - 2020-09-05 20:57 +0100
    Re: Where did Microsoft go wrong? Bob Eager <news0073@eager.cx> - 2020-09-05 20:40 +0000
    Re: Where did Microsoft go wrong? Ahem A Rivet's Shot <steveo@eircom.net> - 2020-09-05 22:44 +0100
      Re: Where did Microsoft go wrong? Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2020-09-08 21:10 +0000
    Re: Where did Microsoft go wrong? Jorgen Grahn <grahn+nntp@snipabacken.se> - 2020-09-05 22:50 +0000
      Re: Where did Microsoft go wrong? J. Clarke <jclarke.873638@gmail.com> - 2020-09-05 19:35 -0400
        Re: Where did Microsoft go wrong? Ahem A Rivet's Shot <steveo@eircom.net> - 2020-09-06 07:12 +0100
          Re: Where did Microsoft go wrong? Quadibloc <jsavard@ecn.ab.ca> - 2021-07-02 09:36 -0700
            Re: Where did Microsoft go wrong? gareth evans <headstone255@yahoo.com> - 2021-07-02 19:05 +0100
              Re: Where did Microsoft go wrong? Ahem A Rivet's Shot <steveo@eircom.net> - 2021-07-02 21:44 +0100
                Re: Where did Microsoft go wrong? gareth evans <headstone255@yahoo.com> - 2021-07-02 22:16 +0100
                  Re: Where did Microsoft go wrong? ted@loft.tnolan.com (Ted Nolan <tednolan>) - 2021-07-03 00:57 +0000
                  Re: Where did Microsoft go wrong? Dave Garland <dave.garland@wizinfo.com> - 2021-07-02 23:05 -0500
                Re: Where did Microsoft go wrong? Fred Weigel <fridtjof.martin.weigel@gmail.com> - 2021-07-02 16:03 -0700
                Re: Where did Microsoft go wrong? scott@slp53.sl.home (Scott Lurndal) - 2021-07-03 17:35 +0000
                  Re: Where did Microsoft go wrong? Ahem A Rivet's Shot <steveo@eircom.net> - 2021-07-04 08:35 +0100
            Re: Where did Microsoft go wrong? Peter Flass <peter_flass@yahoo.com> - 2021-07-02 11:28 -0700
              Re: Where did Microsoft go wrong? Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2021-07-02 23:28 +0000
                Re: Where did Microsoft go wrong? gareth evans <headstone255@yahoo.com> - 2021-07-03 10:21 +0100
                  Re: Where did Microsoft go wrong? Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2021-07-03 22:16 +0000
                  Re: Where did Microsoft go wrong? Andreas Kohlbach <ank@spamfence.net> - 2021-07-03 21:07 -0400
                  Re: Where did Microsoft go wrong? Peter Flass <peter_flass@yahoo.com> - 2021-07-03 18:40 -0700
                  Re: Where did Microsoft go wrong? Quadibloc <jsavard@ecn.ab.ca> - 2021-07-06 04:36 -0700
                    Re: Where did Microsoft go wrong? "Kerr-Mudd, John" <admin@127.0.0.1> - 2021-07-06 17:42 +0100
                    Re: Where did Microsoft go wrong? Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2021-07-06 16:52 +0000
              Re: Where did Microsoft go wrong? Quadibloc <jsavard@ecn.ab.ca> - 2021-07-06 04:34 -0700
      Re: Where did Microsoft go wrong? Peter Flass <peter_flass@yahoo.com> - 2020-09-05 17:59 -0700
        Re: Where did Microsoft go wrong? J. Clarke <jclarke.873638@gmail.com> - 2020-09-05 22:07 -0400
          Re: Where did Microsoft go wrong? Peter Flass <peter_flass@yahoo.com> - 2020-09-06 06:30 -0700
            Re: Where did Microsoft go wrong? J. Clarke <jclarke.873638@gmail.com> - 2020-09-06 10:02 -0400
            Re: Where did Microsoft go wrong? Bud Spencer <bud@campo.verano.it> - 2020-09-06 17:47 +0300
        Re: Where did Microsoft go wrong? Jorgen Grahn <grahn+nntp@snipabacken.se> - 2020-09-06 08:12 +0000
          Re: Where did Microsoft go wrong? Johnatan Duck <pato@thunder.local> - 2020-09-11 08:22 +0000
        Re: Where did Microsoft go wrong? gareth evans <headstone255@yahoo.com> - 2020-09-06 16:23 +0100
      Re: Where did Microsoft go wrong? gareth evans <headstone255@yahoo.com> - 2020-09-06 16:19 +0100
        Re: Where did Microsoft go wrong? Bob Eager <news0073@eager.cx> - 2020-09-06 22:51 +0000
    Re: Where did Microsoft go wrong? Joe Pfeiffer <pfeiffer@cs.nmsu.edu> - 2020-09-05 17:01 -0600
      Re: Where did Microsoft go wrong? Bob Eager <news0073@eager.cx> - 2020-09-06 08:23 +0000
        Re: Where did Microsoft go wrong? scott@slp53.sl.home (Scott Lurndal) - 2020-09-06 20:03 +0000
          Re: Where did Microsoft go wrong? Joe Pfeiffer <pfeiffer@cs.nmsu.edu> - 2020-09-06 17:26 -0600
      Re: Where did Microsoft go wrong? gareth evans <headstone255@yahoo.com> - 2020-09-06 16:20 +0100
        Re: Where did Microsoft go wrong? Peter Flass <peter_flass@yahoo.com> - 2020-09-06 11:14 -0700
          Re: Where did Microsoft go wrong? Thomas Koenig <tkoenig@netcologne.de> - 2020-09-07 05:27 +0000
            Re: Where did Microsoft go wrong? Bob Eager <news0073@eager.cx> - 2020-09-07 08:32 +0000
            Re: Where did Microsoft go wrong? gareth evans <headstone255@yahoo.com> - 2020-09-07 11:01 +0100
            Re: Where did Microsoft go wrong? Peter Flass <peter_flass@yahoo.com> - 2020-09-07 10:28 -0700
            Re: Where did Microsoft go wrong? Dan Espen <dan1espen@gmail.com> - 2020-09-07 13:50 -0400
        Re: Where did Microsoft go wrong? Bob Eager <news0073@eager.cx> - 2020-09-06 22:48 +0000
          Re: Where did Microsoft go wrong? gareth evans <headstone255@yahoo.com> - 2020-09-07 10:58 +0100
    Re: Where did Microsoft go wrong? Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-07 17:47 +0100
      Re: Where did Microsoft go wrong? Ahem A Rivet's Shot <steveo@eircom.net> - 2020-09-07 18:02 +0100
        Re: Where did Microsoft go wrong? Bob Eager <news0073@eager.cx> - 2020-09-07 20:57 +0000
      Re: Where did Microsoft go wrong? Dan Espen <dan1espen@gmail.com> - 2020-09-07 13:44 -0400
        Re: Where did Microsoft go wrong? Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-08 00:05 +0100
          Re: Where did Microsoft go wrong? Dan Espen <dan1espen@gmail.com> - 2020-09-07 19:40 -0400
            Re: Where did Microsoft go wrong? Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-08 01:33 +0100
              Re: Where did Microsoft go wrong? Dan Espen <dan1espen@gmail.com> - 2020-09-07 22:39 -0400
                Re: Where did Microsoft go wrong? Bob Eager <news0073@eager.cx> - 2020-09-08 11:01 +0000
                Re: Where did Microsoft go wrong? Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-08 14:42 +0100
                  Re: Where did Microsoft go wrong? Dan Espen <dan1espen@gmail.com> - 2020-09-08 10:28 -0400
                    Re: Where did Microsoft go wrong? Peter Flass <peter_flass@yahoo.com> - 2020-09-08 08:44 -0700
                    Re: Where did Microsoft go wrong? Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-08 16:48 +0100
                      Re: Where did Microsoft go wrong? Dan Espen <dan1espen@gmail.com> - 2020-09-08 12:12 -0400
                        Re: Where did Microsoft go wrong? Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-08 17:33 +0100
                          Re: Where did Microsoft go wrong? Dan Espen <dan1espen@gmail.com> - 2020-09-08 13:47 -0400
                            Re: Where did Microsoft go wrong? Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-08 19:49 +0100
                              Re: Where did Microsoft go wrong? Dan Espen <dan1espen@gmail.com> - 2020-09-08 15:35 -0400
                    Re: Where did Microsoft go wrong? Mike Spencer <mds@bogus.nodomain.nowhere> - 2020-09-08 20:01 -0300
                      Re: Where did Microsoft go wrong? Dan Espen <dan1espen@gmail.com> - 2020-09-08 20:16 -0400
                        Re: Where did Microsoft go wrong? Mike Spencer <mds@bogus.nodomain.nowhere> - 2020-09-09 03:50 -0300
                          Re: Where did Microsoft go wrong? Charlie Gibbs <cgibbs@kltpzyxm.invalid> - 2020-09-09 17:20 +0000
                            Re: Where did Microsoft go wrong? Mike Spencer <mds@bogus.nodomain.nowhere> - 2020-09-10 03:29 -0300
              Re: Where did Microsoft go wrong? Joe Pfeiffer <pfeiffer@cs.nmsu.edu> - 2020-09-08 09:20 -0600
                Re: Where did Microsoft go wrong? scott@slp53.sl.home (Scott Lurndal) - 2020-09-08 15:39 +0000
                  Re: Where did Microsoft go wrong? Joe Pfeiffer <pfeiffer@cs.nmsu.edu> - 2020-09-08 12:01 -0600
                  Re: Where did Microsoft go wrong? Anssi Saari <as@sci.fi> - 2020-09-09 17:59 +0300
                Re: Where did Microsoft go wrong? Dan Espen <dan1espen@gmail.com> - 2020-09-08 11:54 -0400
                  Re: Where did Microsoft go wrong? Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-08 17:18 +0100
                    Re: Where did Microsoft go wrong? gareth evans <headstone255@yahoo.com> - 2020-09-08 19:44 +0100
            Re: ZFS, Where did Microsoft go wrong? John Levine <johnl@taugh.com> - 2020-09-08 01:27 +0000
              Re: ZFS, Where did Microsoft go wrong? Dan Espen <dan1espen@gmail.com> - 2020-09-07 22:54 -0400
                Re: ZFS, Where did Microsoft go wrong? Peter Flass <peter_flass@yahoo.com> - 2020-09-08 08:44 -0700
                  Re: ZFS, Where did Microsoft go wrong? Niklas Karlsson <anksil@yahoo.se> - 2020-09-08 16:01 +0000
                    Re: ZFS, Where did Microsoft go wrong? Peter Flass <peter_flass@yahoo.com> - 2020-09-08 09:33 -0700
              Re: ZFS, Where did Microsoft go wrong? Peter Flass <peter_flass@yahoo.com> - 2020-09-08 08:44 -0700
                Re: ZFS, Where did Microsoft go wrong? Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-08 17:00 +0100
                  Re: ZFS, Where did Microsoft go wrong? Peter Flass <peter_flass@yahoo.com> - 2020-09-08 09:33 -0700
                    Re: ZFS, -> The Great Rootshow Dan Espen <dan1espen@gmail.com> - 2020-09-08 13:41 -0400
                      Re: ZFS, -> The Great Rootshow Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-08 19:37 +0100
                        Re: ZFS, -> The Great Rootshow Dan Espen <dan1espen@gmail.com> - 2020-09-08 15:31 -0400
                          Re: ZFS, -> The Great Rootshow Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-08 22:54 +0100
                            Re: ZFS, -> WARNING back top systemd Dan Espen <dan1espen@gmail.com> - 2020-09-08 18:09 -0400
                              Re: ZFS, -> WARNING back top systemd Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-08 23:18 +0100
                                Re: ZFS, -> WARNING back top systemd Dan Espen <dan1espen@gmail.com> - 2020-09-08 20:10 -0400
                                  Re: ZFS, -> WARNING back top systemd Bob Eager <news0073@eager.cx> - 2020-09-09 09:41 +0000
                                    Re: ZFS, -> WARNING back top systemd Dan Espen <dan1espen@gmail.com> - 2020-09-09 08:34 -0400
                                      Re: ZFS, -> WARNING back top systemd Bob Eager <news0073@eager.cx> - 2020-09-09 12:43 +0000
                                  Re: ZFS, -> WARNING back top systemd Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-09 13:29 +0100
                                    Re: ZFS, -> WARNING back top systemd Dan Espen <dan1espen@gmail.com> - 2020-09-09 08:44 -0400
                                Re: ZFS, -> WARNING back top systemd scott@slp53.sl.home (Scott Lurndal) - 2020-09-09 18:31 +0000
                                  Re: ZFS, -> WARNING back top systemd Dan Espen <dan1espen@gmail.com> - 2020-09-09 15:49 -0400
                                    Re: ZFS, -> WARNING back top systemd Niklas Karlsson <anksil@yahoo.se> - 2020-09-09 20:41 +0000
                                      Re: ZFS, -> WARNING back top systemd Dan Espen <dan1espen@gmail.com> - 2020-09-09 17:32 -0400
                                        Re: ZFS, -> WARNING back top systemd Thomas Koenig <tkoenig@netcologne.de> - 2020-09-10 05:17 +0000
                                          Re: ZFS, -> WARNING back top systemd gareth evans <headstone255@yahoo.com> - 2020-09-10 12:15 +0100
                                          Re: ZFS, -> WARNING back top systemd Bob Eager <news0073@eager.cx> - 2020-09-10 11:51 +0000
                                            Re: ZFS, -> WARNING back top systemd Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-10 14:40 +0100
                                              Re: ZFS, -> WARNING back top systemd gareth evans <headstone255@yahoo.com> - 2020-09-10 17:39 +0100
                                                Re: ZFS, -> WARNING back top systemd Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-10 18:03 +0100
                                      Re: ZFS, -> WARNING back top systemd scott@slp53.sl.home (Scott Lurndal) - 2020-09-10 15:45 +0000
                                        Re: ZFS, -> WARNING back top systemd Dan Espen <dan1espen@gmail.com> - 2020-09-10 12:02 -0400
                                          Re: ZFS, -> WARNING back top systemd Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-10 17:35 +0100
                                  Re: ZFS, -> WARNING back top systemd Ahem A Rivet's Shot <steveo@eircom.net> - 2020-09-09 21:05 +0100
                                  Re: ZFS, -> WARNING back top systemd Peter Flass <peter_flass@yahoo.com> - 2020-09-09 16:55 -0700
                                    Re: ZFS, -> WARNING back top systemd scott@slp53.sl.home (Scott Lurndal) - 2020-09-10 15:48 +0000
            Re: Where did Microsoft go wrong? scott@slp53.sl.home (Scott Lurndal) - 2020-09-08 15:38 +0000
    Re: Where did Microsoft go wrong? usenet@only.tnx (Questor) - 2020-09-08 02:29 +0000
    Re: Where did Microsoft go wrong? Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-08 17:53 +0100
      Re: Where did Microsoft go wrong? J. Clarke <jclarke.873638@gmail.com> - 2020-09-08 17:43 -0400
        Re: Where did Microsoft go wrong? Chris <xxx.syseng.yyy@gfsys.co.uk> - 2020-09-08 23:00 +0100
    Re: Where did Microsoft go wrong? Robin Vowels <robin.vowels@gmail.com> - 2021-07-02 07:10 -0700
      Re: Where did Microsoft go wrong? gareth evans <headstone255@yahoo.com> - 2021-07-02 16:32 +0100
        Re: Where did Microsoft go wrong? Robin Vowels <robin.vowels@gmail.com> - 2021-07-03 04:10 -0700
      Re: Where did Microsoft go wrong? Robin Vowels <robin.vowels@gmail.com> - 2021-07-08 01:01 -0700
        Re: Where did Microsoft go wrong? "Kerr-Mudd, John" <admin@127.0.0.1> - 2021-07-08 11:36 +0100
          Re: Where did Microsoft go wrong? Rich Alderson <news@alderson.users.panix.com> - 2021-07-08 16:01 -0400
            Re: Where did Microsoft go wrong? Peter Flass <peter_flass@yahoo.com> - 2021-07-08 15:01 -0700
              Re: Where did Microsoft go wrong? scott@slp53.sl.home (Scott Lurndal) - 2021-07-08 22:45 +0000
            Re: Where did Microsoft go wrong? Dan Espen <dan1espen@gmail.com> - 2021-07-08 19:30 -0400
            Re: Where did Microsoft go wrong? Dave Garland <dave.garland@wizinfo.com> - 2021-07-08 21:02 -0500
        Re: Where did Microsoft go wrong? ted@loft.tnolan.com (Ted Nolan <tednolan>) - 2021-07-08 12:09 +0000
          Re: Where did Microsoft go wrong? Andy Burns <usenet@andyburns.uk> - 2021-07-08 13:44 +0100
            Re: Where did Microsoft go wrong? ted@loft.tnolan.com (Ted Nolan <tednolan>) - 2021-07-08 12:49 +0000
              Re: Where did Microsoft go wrong? Andy Burns <usenet@andyburns.uk> - 2021-07-08 13:53 +0100
                Re: Where did Microsoft go wrong? Ahem A Rivet's Shot <steveo@eircom.net> - 2021-07-08 14:17 +0100
                  Re: Where did Microsoft go wrong? Andy Burns <usenet@andyburns.uk> - 2021-07-08 22:00 +0100
        Re: Where did Microsoft go wrong? scott@slp53.sl.home (Scott Lurndal) - 2021-07-08 13:56 +0000
      Re: Where did Microsoft go wrong? Daniel Jendinger <hexabit@daniels-mac-mini2.local> - 2021-09-06 18:56 +0000
        Re: Where did Microsoft go wrong? J. Clarke <jclarke.873638@gmail.com> - 2021-09-06 17:27 -0400
        Re: Where did Microsoft go wrong? gareth evans <headstone255@yahoo.com> - 2021-09-06 22:30 +0100
    Re: Where did Microsoft go wrong? Quadibloc <jsavard@ecn.ab.ca> - 2021-07-02 09:29 -0700

Page 5 of 8 — ← Prev page 1 2 3 4 [5] 6 7 8  Next page →

#213895 — Re: ZFS, Where did Microsoft go wrong?

John Levine <johnl@taugh.com> writes:

> In article <rj6ge8$jkq$1@dont-email.me>,
> Dan Espen  <dan1espen@gmail.com> wrote:
>>> However, from a software engineering point of FreeBSD looks far
>>> more lean in it's design and has other advantages, such as a native
>>> zfs file system and yes, no systemd.
>>
>>Not sure why zfs is a big winner:
>>
>>  ZFS is a combined file system and logical volume manager designed by
>>  Sun Microsystems. ZFS is scalable, and includes extensive protection
>>  against data corruption, support for high storage capacities,
>
> ZFS is great. It handles all the complex disk management in a coherent
> way. It handles combinations of RAID and disk mirroring and SSD caches
> and and lets you efficiently create lots of logical partitions sharing
> a pool of space, but with each partition having different compression,
> filename rules, quotas, and so forth. It also lets you make snapshots
> of a partition's state and efficiently roll forward and back which
> makes backup and restore a lot easier.
>
> My FreeBSD server has four disks configured as two mirrored pairs in
> one pool, with 58 logical partitions so I can manage each user or each
> web site separately.

Yeah, sorry, I was speaking with the home user blinders on.
For my own use I have no use for an LVM and I disabled the one
Fedora likes to turn on.

I've got the system on a 128G SSD and /home on a 1TB SSD.
1TB seemed like a high storage capacity, but with many users
I guess that's a drop in the bucket.  I'm not sure where
my storage would exceed system capacity.

Backups are /home only to 2 USB sticks.  I don't need to backup the
system, if I lost it, I'd just get a new disk and reinstall.

Fedora is pushing xfs, not sure how it's diffrent than ext4 but
I get by with ext4 fine.


-- 
Dan Espen

[toc] | [prev] | [next] | [standalone]

#213929 — Re: ZFS, Where did Microsoft go wrong?

Dan Espen <dan1espen@gmail.com> wrote:
> John Levine <johnl@taugh.com> writes:
> 
>> In article <rj6ge8$jkq$1@dont-email.me>,
>> Dan Espen  <dan1espen@gmail.com> wrote:
>>>> However, from a software engineering point of FreeBSD looks far
>>>> more lean in it's design and has other advantages, such as a native
>>>> zfs file system and yes, no systemd.
>>> 
>>> Not sure why zfs is a big winner:
>>> 
>>> ZFS is a combined file system and logical volume manager designed by
>>> Sun Microsystems. ZFS is scalable, and includes extensive protection
>>> against data corruption, support for high storage capacities,
>> 
>> ZFS is great. It handles all the complex disk management in a coherent
>> way. It handles combinations of RAID and disk mirroring and SSD caches
>> and and lets you efficiently create lots of logical partitions sharing
>> a pool of space, but with each partition having different compression,
>> filename rules, quotas, and so forth. It also lets you make snapshots
>> of a partition's state and efficiently roll forward and back which
>> makes backup and restore a lot easier.
>> 
>> My FreeBSD server has four disks configured as two mirrored pairs in
>> one pool, with 58 logical partitions so I can manage each user or each
>> web site separately.
> 
> Yeah, sorry, I was speaking with the home user blinders on.
> For my own use I have no use for an LVM and I disabled the one
> Fedora likes to turn on.

I tried it, because Fedora wanted to enable it by default. It caused so
much trouble I think I finally had to wipe the disk. At that point I
installed Ubuntu which, although not perfect, seems to be a better fit for
me. The thing is, I was familiar with a lot of LVM from using DFSMS on zOS,
and I still had trouble.

> 
> I've got the system on a 128G SSD and /home on a 1TB SSD.
> 1TB seemed like a high storage capacity, but with many users
> I guess that's a drop in the bucket.  I'm not sure where
> my storage would exceed system capacity.
> 
> Backups are /home only to 2 USB sticks.  I don't need to backup the
> system, if I lost it, I'd just get a new disk and reinstall.
> 
> Fedora is pushing xfs, not sure how it's diffrent than ext4 but
> I get by with ext4 fine.
> 

Works for me, too.

-- 
Pete

[toc] | [prev] | [next] | [standalone]

#213935 — Re: ZFS, Where did Microsoft go wrong?

On 2020-09-08, Peter Flass <peter_flass@yahoo.com> wrote:
> Dan Espen <dan1espen@gmail.com> wrote:
>> Yeah, sorry, I was speaking with the home user blinders on.
>> For my own use I have no use for an LVM and I disabled the one
>> Fedora likes to turn on.
>
> I tried it, because Fedora wanted to enable it by default. It caused so
> much trouble I think I finally had to wipe the disk. At that point I
> installed Ubuntu which, although not perfect, seems to be a better fit for
> me. The thing is, I was familiar with a lot of LVM from using DFSMS on zOS,
> and I still had trouble.

I'm surprised. LVM on Linux has always been smooth sailing for me. Maybe
it gets dicier if you have to shrink a volume and file system, which
I've never really had to do.

LVM is convenient for me even at home, since I run one or more Linux
VMs. Should I run out of space, I can just hang another virtual disk on
it, add it to the volume group and then grow the LV and the file system
in one swell foop. Always worked nicely.

Niklas
-- 
Part of the survival techniques of driving in Massachusetts is to
allow the car to roll back an inch so that the driver behind you
thinks you're a dumb broad who can't drive and stays 5' away.
                       -- Barb Huizenga

[toc] | [prev] | [next] | [standalone]

#213941 — Re: ZFS, Where did Microsoft go wrong?

Niklas Karlsson <anksil@yahoo.se> wrote:
> On 2020-09-08, Peter Flass <peter_flass@yahoo.com> wrote:
>> Dan Espen <dan1espen@gmail.com> wrote:
>>> Yeah, sorry, I was speaking with the home user blinders on.
>>> For my own use I have no use for an LVM and I disabled the one
>>> Fedora likes to turn on.
>> 
>> I tried it, because Fedora wanted to enable it by default. It caused so
>> much trouble I think I finally had to wipe the disk. At that point I
>> installed Ubuntu which, although not perfect, seems to be a better fit for
>> me. The thing is, I was familiar with a lot of LVM from using DFSMS on zOS,
>> and I still had trouble.
> 
> I'm surprised. LVM on Linux has always been smooth sailing for me. Maybe
> it gets dicier if you have to shrink a volume and file system, which
> I've never really had to do.

Maybe that was it. I think I had a full-disk LV and wanted to partition it.
I recall having to edit the LVM definitions by hand to get the system to
work at all, and at that point I decided it wasn’t worth it.

> 
> LVM is convenient for me even at home, since I run one or more Linux
> VMs. Should I run out of space, I can just hang another virtual disk on
> it, add it to the volume group and then grow the LV and the file system
> in one swell foop. Always worked nicely.
> 
> Niklas



-- 
Pete

[toc] | [prev] | [next] | [standalone]

#213928 — Re: ZFS, Where did Microsoft go wrong?

John Levine <johnl@taugh.com> wrote:
> In article <rj6ge8$jkq$1@dont-email.me>,
> Dan Espen  <dan1espen@gmail.com> wrote:
>>> However, from a software engineering point of FreeBSD looks far
>>> more lean in it's design and has other advantages, such as a native
>>> zfs file system and yes, no systemd.
>> 
>> Not sure why zfs is a big winner:
>> 
>> ZFS is a combined file system and logical volume manager designed by
>> Sun Microsystems. ZFS is scalable, and includes extensive protection
>> against data corruption, support for high storage capacities,
> 
> ZFS is great. It handles all the complex disk management in a coherent
> way. It handles combinations of RAID and disk mirroring and SSD caches
> and and lets you efficiently create lots of logical partitions sharing
> a pool of space, but with each partition having different compression,
> filename rules, quotas, and so forth. It also lets you make snapshots
> of a partition's state and efficiently roll forward and back which
> makes backup and restore a lot easier.
> 
> My FreeBSD server has four disks configured as two mirrored pairs in
> one pool, with 58 logical partitions so I can manage each user or each
> web site separately.
> 
> I happen to find FreeBSD's rc.d startup and shutdown scripts quite
> adequate but I realize I'm not going to win arguments with people who
> like systemd.
> 

I’m agnostic. I just want my system to WORK.

-- 
Pete

[toc] | [prev] | [next] | [standalone]

#213934 — Re: ZFS, Where did Microsoft go wrong?

On 09/08/20 16:44, Peter Flass wrote:
> John Levine<johnl@taugh.com>  wrote:
>> In article<rj6ge8$jkq$1@dont-email.me>,
>> Dan Espen<dan1espen@gmail.com>  wrote:
>>>> However, from a software engineering point of FreeBSD looks far
>>>> more lean in it's design and has other advantages, such as a native
>>>> zfs file system and yes, no systemd.
>>>
>>> Not sure why zfs is a big winner:
>>>
>>> ZFS is a combined file system and logical volume manager designed by
>>> Sun Microsystems. ZFS is scalable, and includes extensive protection
>>> against data corruption, support for high storage capacities,
>>
>> ZFS is great. It handles all the complex disk management in a coherent
>> way. It handles combinations of RAID and disk mirroring and SSD caches
>> and and lets you efficiently create lots of logical partitions sharing
>> a pool of space, but with each partition having different compression,
>> filename rules, quotas, and so forth. It also lets you make snapshots
>> of a partition's state and efficiently roll forward and back which
>> makes backup and restore a lot easier.
>>
>> My FreeBSD server has four disks configured as two mirrored pairs in
>> one pool, with 58 logical partitions so I can manage each user or each
>> web site separately.
>>
>> I happen to find FreeBSD's rc.d startup and shutdown scripts quite
>> adequate but I realize I'm not going to win arguments with people who
>> like systemd.
>>
>
> I’m agnostic. I just want my system to WORK.
>

Ditto, and when something does go wrong, not have layers and layers of
opaque cross dependencies to debug it all...

Chris

[toc] | [prev] | [next] | [standalone]

#213940 — Re: ZFS, Where did Microsoft go wrong?

Chris <xxx.syseng.yyy@gfsys.co.uk> wrote:
> On 09/08/20 16:44, Peter Flass wrote:
>> John Levine<johnl@taugh.com>  wrote:
>>> In article<rj6ge8$jkq$1@dont-email.me>,
>>> Dan Espen<dan1espen@gmail.com>  wrote:
>>>>> However, from a software engineering point of FreeBSD looks far
>>>>> more lean in it's design and has other advantages, such as a native
>>>>> zfs file system and yes, no systemd.
>>>> 
>>>> Not sure why zfs is a big winner:
>>>> 
>>>> ZFS is a combined file system and logical volume manager designed by
>>>> Sun Microsystems. ZFS is scalable, and includes extensive protection
>>>> against data corruption, support for high storage capacities,
>>> 
>>> ZFS is great. It handles all the complex disk management in a coherent
>>> way. It handles combinations of RAID and disk mirroring and SSD caches
>>> and and lets you efficiently create lots of logical partitions sharing
>>> a pool of space, but with each partition having different compression,
>>> filename rules, quotas, and so forth. It also lets you make snapshots
>>> of a partition's state and efficiently roll forward and back which
>>> makes backup and restore a lot easier.
>>> 
>>> My FreeBSD server has four disks configured as two mirrored pairs in
>>> one pool, with 58 logical partitions so I can manage each user or each
>>> web site separately.
>>> 
>>> I happen to find FreeBSD's rc.d startup and shutdown scripts quite
>>> adequate but I realize I'm not going to win arguments with people who
>>> like systemd.
>>> 
>> 
>> I’m agnostic. I just want my system to WORK.
>> 
> 
> Ditto, and when something does go wrong, not have layers and layers of
> opaque cross dependencies to debug it all...

Yes. The days of CP/M users “hacking the BIOS”, or whatever it was, for fun
are over for most of us. I mostly use my computer as a tool, although I use
it to develop software. I want the system to fade into the background and
not need constant upgrades and tweaks.

I know some people enjoy customizing and playing with their systems, but
I’ve done too many installs to bother with things like custom fonts and
wallpaper. I have a set of tools that work and just stick with them.
> 
> Chris
> 



-- 
Pete

[toc] | [prev] | [next] | [standalone]

#213944 — Re: ZFS, -> The Great Rootshow

Peter Flass <peter_flass@yahoo.com> writes:

> I’ve done too many installs to bother with things like custom fonts and
> wallpaper.

Speaking of wallpaper...

Can this develop into another war?  I don't see how.

Just folklore.

I'm a big fan of an image on the root.
But then my common mode of operation is to iconify a window then switch
to another window.  So I see the root a lot.

I want an image there, but I've never found an image that I'd want to
stare at day after day.

So, I did this at work and it was a pretty popular feature of the Sun
workstation users that used the environment I established.

I wanted attractive images but I didn't want to get bored.  So I
endeavored to download lots of images and have the computer change the
image every hour.  APOD was a good source, but I found images in lots of
places.  I wanted enough images that the viewer would pretty much forget
the image from the last viewing.

Unix has a lot of programs that can put an image on the root but none of
the images fit exactly.  Some needed to be stretched one way, some the
other.  Some actually looked better tiled over the background.

So, I created a file called "root.commands", each line in the file
contained a command that displayed an image in the best way.
I added a comment at the end of the command that categorized the image,
like 'space far', 'flower rose', 'joke', 'Train', 'SciFi'.

Then I wrote a command called 'calc.next' that looked in the file,
and another file, 'current.line'  and computed the next line to display.
Comments in root.commands were ignored, and the count wrapped around
when the end of the file was reached.  That was executed by each user under
control of cron a few minutes before the background change.

Another command 'do.next'  just used 'current.line' to find and execute
the command from root.commands.  That was in everyone's cron too.

So, that worked pretty well until I added a user that got really upset
by some of the images, especially some of the dragon artwork.

So, I updated calc.next to look into the users '.rootshow' file
for include/bypass rules.  This user settled on flowers only.

Then a few more tweaks.  We had employee children visit the site
in their costumes every Halloween.  I modified the "great rootshow"
to show only Halloween images on Halloween or the weekday before
Halloween.

Lastly I auto-generated Fvwm menus using the comments for categories.
That was quite a bit of menus.  There were over a thousand lines
in root commands.  That way a user could look for and display
an image by its category if the mood took him.

Sadly, all those Sun Workstation users are now relegated to
Windows boxes so the Great Rootshow only lives on on my desktop.


-- 
Dan Espen

[toc] | [prev] | [next] | [standalone]

#213950 — Re: ZFS, -> The Great Rootshow

On 09/08/20 18:41, Dan Espen wrote:
> Peter Flass<peter_flass@yahoo.com>  writes:
>
>> I’ve done too many installs to bother with things like custom fonts and
>> wallpaper.
>
> Speaking of wallpaper...
>
> Can this develop into another war?  I don't see how.
>
> Just folklore.
>
> I'm a big fan of an image on the root.
> But then my common mode of operation is to iconify a window then switch
> to another window.  So I see the root a lot.
>
> I want an image there, but I've never found an image that I'd want to
> stare at day after day.
>
> So, I did this at work and it was a pretty popular feature of the Sun
> workstation users that used the environment I established.
>
> I wanted attractive images but I didn't want to get bored.  So I
> endeavored to download lots of images and have the computer change the
> image every hour.  APOD was a good source, but I found images in lots of
> places.  I wanted enough images that the viewer would pretty much forget
> the image from the last viewing.
>
> Unix has a lot of programs that can put an image on the root but none of
> the images fit exactly.  Some needed to be stretched one way, some the
> other.  Some actually looked better tiled over the background.
>
> So, I created a file called "root.commands", each line in the file
> contained a command that displayed an image in the best way.
> I added a comment at the end of the command that categorized the image,
> like 'space far', 'flower rose', 'joke', 'Train', 'SciFi'.
>
> Then I wrote a command called 'calc.next' that looked in the file,
> and another file, 'current.line'  and computed the next line to display.
> Comments in root.commands were ignored, and the count wrapped around
> when the end of the file was reached.  That was executed by each user under
> control of cron a few minutes before the background change.
>
> Another command 'do.next'  just used 'current.line' to find and execute
> the command from root.commands.  That was in everyone's cron too.
>
> So, that worked pretty well until I added a user that got really upset
> by some of the images, especially some of the dragon artwork.
>
> So, I updated calc.next to look into the users '.rootshow' file
> for include/bypass rules.  This user settled on flowers only.
>
> Then a few more tweaks.  We had employee children visit the site
> in their costumes every Halloween.  I modified the "great rootshow"
> to show only Halloween images on Halloween or the weekday before
> Halloween.
>
> Lastly I auto-generated Fvwm menus using the comments for categories.
> That was quite a bit of menus.  There were over a thousand lines
> in root commands.  That way a user could look for and display
> an image by its category if the mood took him.
>
> Sadly, all those Sun Workstation users are now relegated to
> Windows boxes so the Great Rootshow only lives on on my desktop.
>
>

I always liked the old cde ankh pattern wallpaper on my old
workstations and use that to this day on the current Sol 10
machine. Scaled the pattern using gnu convert, changed the
colour to a cool green, restful to the eye and tiled that across
the screen. Thing is that the pattern looks almost random
from a distance, but look carefully to see the detail...

Chris

[toc] | [prev] | [next] | [standalone]

#213965 — Re: ZFS, -> The Great Rootshow

Chris <xxx.syseng.yyy@gfsys.co.uk> writes:

> I always liked the old cde ankh pattern wallpaper on my old
> workstations and use that to this day on the current Sol 10
> machine. Scaled the pattern using gnu convert, changed the
> colour to a cool green, restful to the eye and tiled that across
> the screen. Thing is that the pattern looks almost random
> from a distance, but look carefully to see the detail...

I believe I have 2 Solaris backgrounds in the root show.
The fishtank and 3d letters spelling SUN.

I mostly use xli (a version of xloadimage) and do the scaling during display.


-- 
Dan Espen

[toc] | [prev] | [next] | [standalone]

#213981 — Re: ZFS, -> The Great Rootshow

On 09/08/20 20:31, Dan Espen wrote:
> Chris<xxx.syseng.yyy@gfsys.co.uk>  writes:
>
>> I always liked the old cde ankh pattern wallpaper on my old
>> workstations and use that to this day on the current Sol 10
>> machine. Scaled the pattern using gnu convert, changed the
>> colour to a cool green, restful to the eye and tiled that across
>> the screen. Thing is that the pattern looks almost random
>> from a distance, but look carefully to see the detail...
>
> I believe I have 2 Solaris backgrounds in the root show.
> The fishtank and 3d letters spelling SUN.
>
> I mostly use xli (a version of xloadimage) and do the scaling during display.
>
>

Whatever, but getting back to systemd, was thinking about how it
works internally, since one of the problems it attempts to solve,
apparently, is the startup of various processes in the correct
order and with dependencies. For example, you don't want
to start up the ntp daemon  before the networking stack
is up and online.

That means it must have an internal priority list for the
things it's responsible for and uses that initially to start
the  lot, but sleeping. Checks back every so often to get
status, so knows when other processes are ready be woken up
and run.

Various ways to do that, but an interesting design problem
none the less...

Chris

[toc] | [prev] | [next] | [standalone]

#213983 — Re: ZFS, -> WARNING back top systemd

Chris <xxx.syseng.yyy@gfsys.co.uk> writes:

> On 09/08/20 20:31, Dan Espen wrote:
>> Chris<xxx.syseng.yyy@gfsys.co.uk>  writes:
>>
>>> I always liked the old cde ankh pattern wallpaper on my old
>>> workstations and use that to this day on the current Sol 10
>>> machine. Scaled the pattern using gnu convert, changed the
>>> colour to a cool green, restful to the eye and tiled that across
>>> the screen. Thing is that the pattern looks almost random
>>> from a distance, but look carefully to see the detail...
>>
>> I believe I have 2 Solaris backgrounds in the root show.
>> The fishtank and 3d letters spelling SUN.
>>
>> I mostly use xli (a version of xloadimage) and do the scaling during display.
>>
>>
>
> Whatever, but getting back to systemd, was thinking about how it
> works internally, since one of the problems it attempts to solve,
> apparently, is the startup of various processes in the correct
> order and with dependencies. For example, you don't want
> to start up the ntp daemon  before the networking stack
> is up and online.
>
> That means it must have an internal priority list for the
> things it's responsible for and uses that initially to start
> the  lot, but sleeping. Checks back every so often to get
> status, so knows when other processes are ready be woken up
> and run.
>
> Various ways to do that, but an interesting design problem
> none the less...

Not sure what you mean by internal priority list.

All of that info is in the .service files.
There's no hard coded logic about services in systemd.

In short what the systemd developers did was look at
everything all those custom designed scripts and soft links were doing
and abstracted it all into the .service files.  You can tell it's a
clean design because each service has only 1 file with just a few
entries being needed for each service.

-- 
Dan Espen

[toc] | [prev] | [next] | [standalone]

#213984 — Re: ZFS, -> WARNING back top systemd

On 09/08/20 23:09, Dan Espen wrote:
> Chris<xxx.syseng.yyy@gfsys.co.uk>  writes:
>
>> On 09/08/20 20:31, Dan Espen wrote:
>>> Chris<xxx.syseng.yyy@gfsys.co.uk>   writes:
>>>
>>>> I always liked the old cde ankh pattern wallpaper on my old
>>>> workstations and use that to this day on the current Sol 10
>>>> machine. Scaled the pattern using gnu convert, changed the
>>>> colour to a cool green, restful to the eye and tiled that across
>>>> the screen. Thing is that the pattern looks almost random
>>>> from a distance, but look carefully to see the detail...
>>>
>>> I believe I have 2 Solaris backgrounds in the root show.
>>> The fishtank and 3d letters spelling SUN.
>>>
>>> I mostly use xli (a version of xloadimage) and do the scaling during display.
>>>
>>>
>>
>> Whatever, but getting back to systemd, was thinking about how it
>> works internally, since one of the problems it attempts to solve,
>> apparently, is the startup of various processes in the correct
>> order and with dependencies. For example, you don't want
>> to start up the ntp daemon  before the networking stack
>> is up and online.
>>
>> That means it must have an internal priority list for the
>> things it's responsible for and uses that initially to start
>> the  lot, but sleeping. Checks back every so often to get
>> status, so knows when other processes are ready be woken up
>> and run.
>>
>> Various ways to do that, but an interesting design problem
>> none the less...
>
> Not sure what you mean by internal priority list.

Some data file that specifies the dependencies and order in
which the various services are started up.

>
> All of that info is in the .service files.
> There's no hard coded logic about services in systemd.

I wouldn't expect it to. Minimally, all it needs is an ordered
list of processes and their dependencies.

>
> In short what the systemd developers did was look at
> everything all those custom designed scripts and soft links were doing
> and abstracted it all into the .service files.  You can tell it's a
> clean design because each service has only 1 file with just a few
> entries being needed for each service.
>

That's a different issue to the priority problem, which I understood
was one of the things that it was designed to address. Maybe not...

Chris

[toc] | [prev] | [next] | [standalone]

#213992 — Re: ZFS, -> WARNING back top systemd

Chris <xxx.syseng.yyy@gfsys.co.uk> writes:

> On 09/08/20 23:09, Dan Espen wrote:
>> Chris<xxx.syseng.yyy@gfsys.co.uk>  writes:
>>
>>> On 09/08/20 20:31, Dan Espen wrote:
>>>> Chris<xxx.syseng.yyy@gfsys.co.uk>   writes:
>>>>
>>>>> I always liked the old cde ankh pattern wallpaper on my old
>>>>> workstations and use that to this day on the current Sol 10
>>>>> machine. Scaled the pattern using gnu convert, changed the
>>>>> colour to a cool green, restful to the eye and tiled that across
>>>>> the screen. Thing is that the pattern looks almost random
>>>>> from a distance, but look carefully to see the detail...
>>>>
>>>> I believe I have 2 Solaris backgrounds in the root show.
>>>> The fishtank and 3d letters spelling SUN.
>>>>
>>>> I mostly use xli (a version of xloadimage) and do the scaling during display.
>>>>
>>>>
>>>
>>> Whatever, but getting back to systemd, was thinking about how it
>>> works internally, since one of the problems it attempts to solve,
>>> apparently, is the startup of various processes in the correct
>>> order and with dependencies. For example, you don't want
>>> to start up the ntp daemon  before the networking stack
>>> is up and online.
>>>
>>> That means it must have an internal priority list for the
>>> things it's responsible for and uses that initially to start
>>> the  lot, but sleeping. Checks back every so often to get
>>> status, so knows when other processes are ready be woken up
>>> and run.
>>>
>>> Various ways to do that, but an interesting design problem
>>> none the less...
>>
>> Not sure what you mean by internal priority list.
>
> Some data file that specifies the dependencies and order in
> which the various services are started up.
>
>>
>> All of that info is in the .service files.
>> There's no hard coded logic about services in systemd.
>
> I wouldn't expect it to. Minimally, all it needs is an ordered
> list of processes and their dependencies.

It's not really an ordered list;

Here is the NFS server .service file:

> cat /usr/lib/systemd/system/nfs-server.service
[Unit]
Description=NFS server and services
DefaultDependencies=no
Requires=network.target proc-fs-nfsd.mount
Requires=nfs-mountd.service
Wants=rpcbind.socket network-online.target
Wants=rpc-statd.service nfs-idmapd.service
Wants=rpc-statd-notify.service
Wants=nfsdcld.service

After=network-online.target local-fs.target
After=proc-fs-nfsd.mount rpcbind.socket nfs-mountd.service
After=nfs-idmapd.service rpc-statd.service
After=nfsdcld.service
Before=rpc-statd-notify.service

# GSS services dependencies and ordering
Wants=auth-rpcgss-module.service
After=rpc-gssd.service gssproxy.service rpc-svcgssd.service

[Service]
Type=oneshot
RemainAfterExit=yes
ExecStartPre=/usr/sbin/exportfs -r
ExecStart=/usr/sbin/rpc.nfsd
ExecStart=-/bin/sh -c 'if systemctl -q is-active gssproxy; then systemctl reload gssproxy ; fi'
ExecStop=/usr/sbin/rpc.nfsd 0
ExecStopPost=/usr/sbin/exportfs -au
ExecStopPost=/usr/sbin/exportfs -f

ExecReload=/usr/sbin/exportfs -r

[Install]
WantedBy=multi-user.target


Notice how it declares what it requires, it's order, and what wants it.

Internally systemd creates a tree structure and works up and down the
tree as needed.

>> In short what the systemd developers did was look at
>> everything all those custom designed scripts and soft links were doing
>> and abstracted it all into the .service files.  You can tell it's a
>> clean design because each service has only 1 file with just a few
>> entries being needed for each service.
>>
>
> That's a different issue to the priority problem, which I understood
> was one of the things that it was designed to address. Maybe not...

The .service files declare interdependence.  (What you're calling
priority.)


-- 
Dan Espen

[toc] | [prev] | [next] | [standalone]

#214002 — Re: ZFS, -> WARNING back top systemd

On Tue, 08 Sep 2020 20:10:12 -0400, Dan Espen wrote:

> Chris <xxx.syseng.yyy@gfsys.co.uk> writes:
> 
>> On 09/08/20 23:09, Dan Espen wrote:
>>> Chris<xxx.syseng.yyy@gfsys.co.uk>  writes:
>>>
>>>> On 09/08/20 20:31, Dan Espen wrote:
>>>>> Chris<xxx.syseng.yyy@gfsys.co.uk>   writes:
>>>>>
>>>>>> I always liked the old cde ankh pattern wallpaper on my old
>>>>>> workstations and use that to this day on the current Sol 10
>>>>>> machine. Scaled the pattern using gnu convert, changed the colour
>>>>>> to a cool green, restful to the eye and tiled that across the
>>>>>> screen. Thing is that the pattern looks almost random from a
>>>>>> distance, but look carefully to see the detail...
>>>>>
>>>>> I believe I have 2 Solaris backgrounds in the root show.
>>>>> The fishtank and 3d letters spelling SUN.
>>>>>
>>>>> I mostly use xli (a version of xloadimage) and do the scaling during
>>>>> display.
>>>>>
>>>>>
>>>>>
>>>> Whatever, but getting back to systemd, was thinking about how it
>>>> works internally, since one of the problems it attempts to solve,
>>>> apparently, is the startup of various processes in the correct order
>>>> and with dependencies. For example, you don't want to start up the
>>>> ntp daemon  before the networking stack is up and online.
>>>>
>>>> That means it must have an internal priority list for the things it's
>>>> responsible for and uses that initially to start the  lot, but
>>>> sleeping. Checks back every so often to get status, so knows when
>>>> other processes are ready be woken up and run.
>>>>
>>>> Various ways to do that, but an interesting design problem none the
>>>> less...
>>>
>>> Not sure what you mean by internal priority list.
>>
>> Some data file that specifies the dependencies and order in which the
>> various services are started up.
>>
>>
>>> All of that info is in the .service files.
>>> There's no hard coded logic about services in systemd.
>>
>> I wouldn't expect it to. Minimally, all it needs is an ordered list of
>> processes and their dependencies.
> 
> It's not really an ordered list;
> 
> Here is the NFS server .service file:
> 
>> cat /usr/lib/systemd/system/nfs-server.service
> [Unit]
> Description=NFS server and services DefaultDependencies=no
> Requires=network.target proc-fs-nfsd.mount Requires=nfs-mountd.service
> Wants=rpcbind.socket network-online.target Wants=rpc-statd.service
> nfs-idmapd.service Wants=rpc-statd-notify.service Wants=nfsdcld.service
> 
> After=network-online.target local-fs.target After=proc-fs-nfsd.mount
> rpcbind.socket nfs-mountd.service After=nfs-idmapd.service
> rpc-statd.service After=nfsdcld.service Before=rpc-statd-notify.service
> 
> # GSS services dependencies and ordering
> Wants=auth-rpcgss-module.service After=rpc-gssd.service gssproxy.service
> rpc-svcgssd.service
> 
> [Service]
> Type=oneshot RemainAfterExit=yes ExecStartPre=/usr/sbin/exportfs -r
> ExecStart=/usr/sbin/rpc.nfsd ExecStart=-/bin/sh -c 'if systemctl -q
> is-active gssproxy; then systemctl reload gssproxy ; fi'
> ExecStop=/usr/sbin/rpc.nfsd 0 ExecStopPost=/usr/sbin/exportfs -au
> ExecStopPost=/usr/sbin/exportfs -f
> 
> ExecReload=/usr/sbin/exportfs -r
> 
> [Install]
> WantedBy=multi-user.target
> 
> 
> Notice how it declares what it requires, it's order, and what wants it.
> 
> Internally systemd creates a tree structure and works up and down the
> tree as needed.
> 
>>> In short what the systemd developers did was look at everything all
>>> those custom designed scripts and soft links were doing and abstracted
>>> it all into the .service files.  You can tell it's a clean design
>>> because each service has only 1 file with just a few entries being
>>> needed for each service.
>>>
>>>
>> That's a different issue to the priority problem, which I understood
>> was one of the things that it was designed to address. Maybe not...
> 
> The .service files declare interdependence.  (What you're calling
> priority.)

No different on FreeBSD really (but without systemd!). A single script, 
which also declares dependencies.

The script starts:

#!/bin/sh

# PROVIDE: nfsd
# REQUIRE: mountcritremote mountd hostname gssd nfsuserd
# KEYWORD: nojail shutdown

. /etc/rc.subr

name="nfsd"
desc="Remote NFS server"
rcvar="nfs_server_enable"
command="/usr/sbin/${name}"


-- 
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

[toc] | [prev] | [next] | [standalone]

#214006 — Re: ZFS, -> WARNING back top systemd

Bob Eager <news0073@eager.cx> writes:
> On Tue, 08 Sep 2020 20:10:12 -0400, Dan Espen wrote:
>
>> Chris <xxx.syseng.yyy@gfsys.co.uk> writes:
>> 
>>> On 09/08/20 23:09, Dan Espen wrote:
>>>> Chris<xxx.syseng.yyy@gfsys.co.uk>  writes:
>>>>
>>>>> On 09/08/20 20:31, Dan Espen wrote:
>>>>>> Chris<xxx.syseng.yyy@gfsys.co.uk>   writes:
>>>>>>
>>>>>>> I always liked the old cde ankh pattern wallpaper on my old
>>>>>>> workstations and use that to this day on the current Sol 10
>>>>>>> machine. Scaled the pattern using gnu convert, changed the colour
>>>>>>> to a cool green, restful to the eye and tiled that across the
>>>>>>> screen. Thing is that the pattern looks almost random from a
>>>>>>> distance, but look carefully to see the detail...
>>>>>>
>>>>>> I believe I have 2 Solaris backgrounds in the root show.
>>>>>> The fishtank and 3d letters spelling SUN.
>>>>>>
>>>>>> I mostly use xli (a version of xloadimage) and do the scaling during
>>>>>> display.
>>>>>>
>>>>>>
>>>>>>
>>>>> Whatever, but getting back to systemd, was thinking about how it
>>>>> works internally, since one of the problems it attempts to solve,
>>>>> apparently, is the startup of various processes in the correct order
>>>>> and with dependencies. For example, you don't want to start up the
>>>>> ntp daemon  before the networking stack is up and online.
>>>>>
>>>>> That means it must have an internal priority list for the things it's
>>>>> responsible for and uses that initially to start the  lot, but
>>>>> sleeping. Checks back every so often to get status, so knows when
>>>>> other processes are ready be woken up and run.
>>>>>
>>>>> Various ways to do that, but an interesting design problem none the
>>>>> less...
>>>>
>>>> Not sure what you mean by internal priority list.
>>>
>>> Some data file that specifies the dependencies and order in which the
>>> various services are started up.
>>>
>>>
>>>> All of that info is in the .service files.
>>>> There's no hard coded logic about services in systemd.
>>>
>>> I wouldn't expect it to. Minimally, all it needs is an ordered list of
>>> processes and their dependencies.
>> 
>> It's not really an ordered list;
>> 
>> Here is the NFS server .service file:
>> 
>>> cat /usr/lib/systemd/system/nfs-server.service
>> [Unit]
>> Description=NFS server and services DefaultDependencies=no
>> Requires=network.target proc-fs-nfsd.mount Requires=nfs-mountd.service
>> Wants=rpcbind.socket network-online.target Wants=rpc-statd.service
>> nfs-idmapd.service Wants=rpc-statd-notify.service Wants=nfsdcld.service
>> 
>> After=network-online.target local-fs.target After=proc-fs-nfsd.mount
>> rpcbind.socket nfs-mountd.service After=nfs-idmapd.service
>> rpc-statd.service After=nfsdcld.service Before=rpc-statd-notify.service
>> 
>> # GSS services dependencies and ordering
>> Wants=auth-rpcgss-module.service After=rpc-gssd.service gssproxy.service
>> rpc-svcgssd.service
>> 
>> [Service]
>> Type=oneshot RemainAfterExit=yes ExecStartPre=/usr/sbin/exportfs -r
>> ExecStart=/usr/sbin/rpc.nfsd ExecStart=-/bin/sh -c 'if systemctl -q
>> is-active gssproxy; then systemctl reload gssproxy ; fi'
>> ExecStop=/usr/sbin/rpc.nfsd 0 ExecStopPost=/usr/sbin/exportfs -au
>> ExecStopPost=/usr/sbin/exportfs -f
>> 
>> ExecReload=/usr/sbin/exportfs -r
>> 
>> [Install]
>> WantedBy=multi-user.target
>> 
>> 
>> Notice how it declares what it requires, it's order, and what wants it.
>> 
>> Internally systemd creates a tree structure and works up and down the
>> tree as needed.
>> 
>>>> In short what the systemd developers did was look at everything all
>>>> those custom designed scripts and soft links were doing and abstracted
>>>> it all into the .service files.  You can tell it's a clean design
>>>> because each service has only 1 file with just a few entries being
>>>> needed for each service.
>>>>
>>>>
>>> That's a different issue to the priority problem, which I understood
>>> was one of the things that it was designed to address. Maybe not...
>> 
>> The .service files declare interdependence.  (What you're calling
>> priority.)
>
> No different on FreeBSD really (but without systemd!). A single script, 
> which also declares dependencies.
>
> The script starts:
>
> #!/bin/sh
>
> # PROVIDE: nfsd
> # REQUIRE: mountcritremote mountd hostname gssd nfsuserd
> # KEYWORD: nojail shutdown
>
> . /etc/rc.subr
>
> name="nfsd"
> desc="Remote NFS server"
> rcvar="nfs_server_enable"
> command="/usr/sbin/${name}"

Actually, that looks pretty good.
Still not as simple as one flat file full of keywords, but at least
the comments and variables convey some of the same info as systemd.

-- 
Dan Espen

[toc] | [prev] | [next] | [standalone]

#214007 — Re: ZFS, -> WARNING back top systemd

On Wed, 09 Sep 2020 08:34:53 -0400, Dan Espen wrote:

> Bob Eager <news0073@eager.cx> writes:
>> On Tue, 08 Sep 2020 20:10:12 -0400, Dan Espen wrote:
>>
>>> Chris <xxx.syseng.yyy@gfsys.co.uk> writes:
>>> 
>>>> On 09/08/20 23:09, Dan Espen wrote:
>>>>> Chris<xxx.syseng.yyy@gfsys.co.uk>  writes:
>>>>>
>>>>>> On 09/08/20 20:31, Dan Espen wrote:
>>>>>>> Chris<xxx.syseng.yyy@gfsys.co.uk>   writes:
>>>>>>>
>>>>>>>> I always liked the old cde ankh pattern wallpaper on my old
>>>>>>>> workstations and use that to this day on the current Sol 10
>>>>>>>> machine. Scaled the pattern using gnu convert, changed the colour
>>>>>>>> to a cool green, restful to the eye and tiled that across the
>>>>>>>> screen. Thing is that the pattern looks almost random from a
>>>>>>>> distance, but look carefully to see the detail...
>>>>>>>
>>>>>>> I believe I have 2 Solaris backgrounds in the root show.
>>>>>>> The fishtank and 3d letters spelling SUN.
>>>>>>>
>>>>>>> I mostly use xli (a version of xloadimage) and do the scaling
>>>>>>> during display.
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>> Whatever, but getting back to systemd, was thinking about how it
>>>>>> works internally, since one of the problems it attempts to solve,
>>>>>> apparently, is the startup of various processes in the correct
>>>>>> order and with dependencies. For example, you don't want to start
>>>>>> up the ntp daemon  before the networking stack is up and online.
>>>>>>
>>>>>> That means it must have an internal priority list for the things
>>>>>> it's responsible for and uses that initially to start the  lot, but
>>>>>> sleeping. Checks back every so often to get status, so knows when
>>>>>> other processes are ready be woken up and run.
>>>>>>
>>>>>> Various ways to do that, but an interesting design problem none the
>>>>>> less...
>>>>>
>>>>> Not sure what you mean by internal priority list.
>>>>
>>>> Some data file that specifies the dependencies and order in which the
>>>> various services are started up.
>>>>
>>>>
>>>>> All of that info is in the .service files.
>>>>> There's no hard coded logic about services in systemd.
>>>>
>>>> I wouldn't expect it to. Minimally, all it needs is an ordered list
>>>> of processes and their dependencies.
>>> 
>>> It's not really an ordered list;
>>> 
>>> Here is the NFS server .service file:
>>> 
>>>> cat /usr/lib/systemd/system/nfs-server.service
>>> [Unit]
>>> Description=NFS server and services DefaultDependencies=no
>>> Requires=network.target proc-fs-nfsd.mount Requires=nfs-mountd.service
>>> Wants=rpcbind.socket network-online.target Wants=rpc-statd.service
>>> nfs-idmapd.service Wants=rpc-statd-notify.service
>>> Wants=nfsdcld.service
>>> 
>>> After=network-online.target local-fs.target After=proc-fs-nfsd.mount
>>> rpcbind.socket nfs-mountd.service After=nfs-idmapd.service
>>> rpc-statd.service After=nfsdcld.service
>>> Before=rpc-statd-notify.service
>>> 
>>> # GSS services dependencies and ordering
>>> Wants=auth-rpcgss-module.service After=rpc-gssd.service
>>> gssproxy.service rpc-svcgssd.service
>>> 
>>> [Service]
>>> Type=oneshot RemainAfterExit=yes ExecStartPre=/usr/sbin/exportfs -r
>>> ExecStart=/usr/sbin/rpc.nfsd ExecStart=-/bin/sh -c 'if systemctl -q
>>> is-active gssproxy; then systemctl reload gssproxy ; fi'
>>> ExecStop=/usr/sbin/rpc.nfsd 0 ExecStopPost=/usr/sbin/exportfs -au
>>> ExecStopPost=/usr/sbin/exportfs -f
>>> 
>>> ExecReload=/usr/sbin/exportfs -r
>>> 
>>> [Install]
>>> WantedBy=multi-user.target
>>> 
>>> 
>>> Notice how it declares what it requires, it's order, and what wants
>>> it.
>>> 
>>> Internally systemd creates a tree structure and works up and down the
>>> tree as needed.
>>> 
>>>>> In short what the systemd developers did was look at everything all
>>>>> those custom designed scripts and soft links were doing and
>>>>> abstracted it all into the .service files.  You can tell it's a
>>>>> clean design because each service has only 1 file with just a few
>>>>> entries being needed for each service.
>>>>>
>>>>>
>>>> That's a different issue to the priority problem, which I understood
>>>> was one of the things that it was designed to address. Maybe not...
>>> 
>>> The .service files declare interdependence.  (What you're calling
>>> priority.)
>>
>> No different on FreeBSD really (but without systemd!). A single script,
>> which also declares dependencies.
>>
>> The script starts:
>>
>> #!/bin/sh
>>
>> # PROVIDE: nfsd # REQUIRE: mountcritremote mountd hostname gssd
>> nfsuserd # KEYWORD: nojail shutdown
>>
>> . /etc/rc.subr
>>
>> name="nfsd"
>> desc="Remote NFS server"
>> rcvar="nfs_server_enable"
>> command="/usr/sbin/${name}"
> 
> Actually, that looks pretty good.
> Still not as simple as one flat file full of keywords, but at least the
> comments and variables convey some of the same info as systemd.

Packages that have daemons automatically install the file, and no editing 
is required except for the one central file that enables it.



-- 
Using UNIX since v6 (1975)...

Use the BIG mirror service in the UK:
 http://www.mirrorservice.org

[toc] | [prev] | [next] | [standalone]

#214005 — Re: ZFS, -> WARNING back top systemd

On 09/09/20 01:10, Dan Espen wrote:
> Chris<xxx.syseng.yyy@gfsys.co.uk>  writes:
>
>> On 09/08/20 23:09, Dan Espen wrote:
>>> Chris<xxx.syseng.yyy@gfsys.co.uk>   writes:
>>>
>>>> On 09/08/20 20:31, Dan Espen wrote:
>>>>> Chris<xxx.syseng.yyy@gfsys.co.uk>    writes:
>>>>>
>>>>>> I always liked the old cde ankh pattern wallpaper on my old
>>>>>> workstations and use that to this day on the current Sol 10
>>>>>> machine. Scaled the pattern using gnu convert, changed the
>>>>>> colour to a cool green, restful to the eye and tiled that across
>>>>>> the screen. Thing is that the pattern looks almost random
>>>>>> from a distance, but look carefully to see the detail...
>>>>>
>>>>> I believe I have 2 Solaris backgrounds in the root show.
>>>>> The fishtank and 3d letters spelling SUN.
>>>>>
>>>>> I mostly use xli (a version of xloadimage) and do the scaling during display.
>>>>>
>>>>>
>>>>
>>>> Whatever, but getting back to systemd, was thinking about how it
>>>> works internally, since one of the problems it attempts to solve,
>>>> apparently, is the startup of various processes in the correct
>>>> order and with dependencies. For example, you don't want
>>>> to start up the ntp daemon  before the networking stack
>>>> is up and online.
>>>>
>>>> That means it must have an internal priority list for the
>>>> things it's responsible for and uses that initially to start
>>>> the  lot, but sleeping. Checks back every so often to get
>>>> status, so knows when other processes are ready be woken up
>>>> and run.
>>>>
>>>> Various ways to do that, but an interesting design problem
>>>> none the less...
>>>
>>> Not sure what you mean by internal priority list.
>>
>> Some data file that specifies the dependencies and order in
>> which the various services are started up.
>>
>>>
>>> All of that info is in the .service files.
>>> There's no hard coded logic about services in systemd.
>>
>> I wouldn't expect it to. Minimally, all it needs is an ordered
>> list of processes and their dependencies.
>
> It's not really an ordered list;
>
> Here is the NFS server .service file:
>
>> cat /usr/lib/systemd/system/nfs-server.service
> [Unit]
> Description=NFS server and services
> DefaultDependencies=no
> Requires=network.target proc-fs-nfsd.mount
> Requires=nfs-mountd.service
> Wants=rpcbind.socket network-online.target
> Wants=rpc-statd.service nfs-idmapd.service
> Wants=rpc-statd-notify.service
> Wants=nfsdcld.service
>
> After=network-online.target local-fs.target
> After=proc-fs-nfsd.mount rpcbind.socket nfs-mountd.service
> After=nfs-idmapd.service rpc-statd.service
> After=nfsdcld.service
> Before=rpc-statd-notify.service
>
> # GSS services dependencies and ordering
> Wants=auth-rpcgss-module.service
> After=rpc-gssd.service gssproxy.service rpc-svcgssd.service
>
> [Service]
> Type=oneshot
> RemainAfterExit=yes
> ExecStartPre=/usr/sbin/exportfs -r
> ExecStart=/usr/sbin/rpc.nfsd
> ExecStart=-/bin/sh -c 'if systemctl -q is-active gssproxy; then systemctl reload gssproxy ; fi'
> ExecStop=/usr/sbin/rpc.nfsd 0
> ExecStopPost=/usr/sbin/exportfs -au
> ExecStopPost=/usr/sbin/exportfs -f
>
> ExecReload=/usr/sbin/exportfs -r
>
> [Install]
> WantedBy=multi-user.target
>
>
> Notice how it declares what it requires, it's order, and what wants it.
>
> Internally systemd creates a tree structure and works up and down the
> tree as needed.
>
>>> In short what the systemd developers did was look at
>>> everything all those custom designed scripts and soft links were doing
>>> and abstracted it all into the .service files.  You can tell it's a
>>> clean design because each service has only 1 file with just a few
>>> entries being needed for each service.
>>>
>>
>> That's a different issue to the priority problem, which I understood
>> was one of the things that it was designed to address. Maybe not...
>
> The .service files declare interdependence.  (What you're calling
> priority.)
>
>

I did find a brief description of the systemd internals here:

https://bpowers.net/weblog/2016/systemd-internals/

If that is clear to you, good luck :-).

It looks very complex for what it is and the syntax is very
wordy. Compare that with Bob's description of the FreeBSD startup
process, which looks quite clear and builds on what was already
there, not a completely new subsystem with it's fingers into
everything.

I had started to think yesterday that systemd might have some redeeming
features, but that has been corrected today...

Chris

[toc] | [prev] | [next] | [standalone]

#214008 — Re: ZFS, -> WARNING back top systemd

Chris <xxx.syseng.yyy@gfsys.co.uk> writes:

> I did find a brief description of the systemd internals here:
>
> https://bpowers.net/weblog/2016/systemd-internals/
>
> If that is clear to you, good luck :-).

It's fairly clear.

It's not information that any user should need, it's a description
of some of the routines in systemd.

> It looks very complex for what it is and the syntax is very
> wordy. Compare that with Bob's description of the FreeBSD startup
> process, which looks quite clear and builds on what was already
> there, not a completely new subsystem with it's fingers into
> everything.
>
> I had started to think yesterday that systemd might have some redeeming
> features, but that has been corrected today...

I hope not by that information above.
It's irrelevant to anyone not trying to read the source code.

For systemd, you want to start with 'man systemd'.
That will lead you to lots of good stuff including a description
of the .service files in 'man systemd-system.conf'.

-- 
Dan Espen

[toc] | [prev] | [next] | [standalone]

#214024 — Re: ZFS, -> WARNING back top systemd

Chris <xxx.syseng.yyy@gfsys.co.uk> writes:
>On 09/08/20 23:09, Dan Espen wrote:

>>
>> In short what the systemd developers did was look at
>> everything all those custom designed scripts and soft links were doing
>> and abstracted it all into the .service files.  You can tell it's a
>> clean design because each service has only 1 file with just a few
>> entries being needed for each service.
>>
>
>That's a different issue to the priority problem, which I understood
>was one of the things that it was designed to address. Maybe not...

The priority system was simple and effective using the standard
System V intialization scripts.   They were executed in collating
order in the default C locale.

So S01xxxx would be started before S93yyyy.

the 'chkconfig' command would enable/disable services at the target runlevel.
(e.g. chkconfig network on)

The 'service' command would start/stop/restart services
(e.g. service network start).

Simple.  Easy.  Maintainable.  Flexible.  Independent.

chkconfig --list

shows all the services and the run-level they're enabled for.

chkconfig --list | grep ":on"

shows all enabled services


For systemd, you need
  'systemctl list-unit-files'
which is completely unintuitive, and on my system has
384 unit files (which far more than the 20 or 30 services
managed on a typical system v init system); most of them
useless for everyday system management.

[toc] | [prev] | [next] | [standalone]

Page 5 of 8 — ← Prev page 1 2 3 4 [5] 6 7 8  Next page →

Back to top | Article view | alt.folklore.computers

csiph-web