|
|
|
|
|
|
| Author |
Message |
Rich Teer
*nix forums Guru
Joined: 21 Feb 2005
Posts: 1038
|
 Posted: Thu Feb 16, 2006 11:45 pm Post subject:
Re: zones, disks & sds slices
|
|
|
On Thu, 16 Feb 2006, Lion-O wrote:
| Quote: | Oops.. Sorry about that  It is indeed one of the things I'm really looking
forward to.
|
ZFS is available TODAY if you install build 28+ (current is build 33)
of Solaris Express. And it is VERY drool-worthy! :-)
--
Rich Teer, SCNA, SCSA, OpenSolaris CAB member
President,
Rite Online Inc.
Voice: +1 (250) 979-1638
URL: http://www.rite-group.com/rich |
|
| Back to top |
|
 |
Lion-O
*nix forums Guru
Joined: 24 Mar 2005
Posts: 346
|
| Posted: Thu Feb 16, 2006 10:40 pm Post subject:
Re: zones, disks & sds slices
|
|
|
| Quote: | Groups and users. And since your root (and any other slice you mount in the
zone) will be picked up as a seperate filesystem...
ufs doesn't support group quotas, other may.
|
Ayups, thanks a bunch for that addition. From time to time I simply keep mixing
up Linux and Solaris experiences. Very annoying.
| Quote: | Or single filesystem in ZFS speak; ZFS will make this all much easier and
flexible. (It allows you to pool all the unused bits whereas separate slices
do not)
|
*drool*
Oops.. Sorry about that It is indeed one of the things I'm really looking
forward to.
--
Groetjes, Peter
..\\ PGP/GPG key: http://www.catslair.org/pubkey.asc |
|
| Back to top |
|
|
Jerome Quelin
*nix forums beginner
Joined: 14 Feb 2006
Posts: 3
|
| Posted: Wed Feb 15, 2006 3:47 pm Post subject:
Re: zones, disks & sds slices
|
|
|
On 06/02/15 03:47 +0100, Bruno Bonfils wrote:
| Quote: | Why ? What is the problem with that ?
bash-3.00# metastat -p
d16 -p d3 -o 16777440 -b 1048576
d3 -m d2 d1 1
d2 1 1 c2d0s7
d1 1 1 c1d0s7
d15 -p d3 -o 12583104 -b 4194304
d14 -p d3 -o 8388768 -b 4194304
d13 -p d3 -o 6291584 -b 2097152
d12 -p d3 -o 5242976 -b 1048576
d11 -p d3 -o 4194368 -b 1048576
d10 -p d3 -o 32 -b 4194304
d1 are d2 are the physical devices, d3 the mirror, and dN where N >=
10 are soft partitions, some of them used as zonepath, others for fs
used only inside non global zones.
|
d'oh! that's better indeed.
(thanks also to Darren Dunham and Rich Teer for this solution)
| Quote: | I use such configuration for few months (~6/, never had problems.
|
thanks also for the feedback.
regards,
jérôme
--
jquelin@mongueurs.net |
|
| Back to top |
|
|
Robert Milkowski
*nix forums addict
Joined: 24 Feb 2005
Posts: 96
|
| Posted: Wed Feb 15, 2006 10:51 am Post subject:
Re: zones, disks & sds slices
|
|
|
Jerome Quelin <jquelin@mongueurs.net> wrote:
| Quote: | hey,
with solaris 10 comes the zones facility. in order to segregate zones,
i'd like to create a filesystem of its own for each of the zones' root
filesystem.
current disks are now 73gb+, but still only 7 slices... if one count /,
/var, swap and metadb, this leaves only 3 slices on the disks.
so i'd like to know how you guys are dealing with this? since zfs is
still not there - and won't be able to be installed on boot disks anyway
at first - i'm stuck with sds...
are you using soft partitions? since i'm using a two-way mirrors, this
would force me to create the following metadevices (for example):
d30 -m d100 d110
d100 1 1 d120
d110 1 1 d130
d120 -p c1t0d0s5 -o 1 -b 8389656
d130 -p c1t1d0s5 -o 1 -b 8389656
==> which is quite clumsy - 5 metadevices for a lone usable metadevice!
so, how are you dealing with this situation?
|
As someone else pointed out if you really need to have different
filesystems (quota) for each zone then SVM Soft Partitions are right
choice.
ZFS - although you won't be able to *boot* from ZFS with first stable release
nothing prevents you to create one large slice with remaining space
on your local disks then do stripe/raidz/mirror/whatever with zfs
and put zones on that pool (within different filesystems if you want).
That's what I'm doing right now with Nevada builds.
--
Robert Milkowski
rmilkowskiDSADAS@wp-sa.pl
http://milek.blogspot.com |
|
| Back to top |
|
|
Bruno Bonfils
*nix forums beginner
Joined: 02 May 2005
Posts: 22
|
| Posted: Wed Feb 15, 2006 2:47 am Post subject:
Re: zones, disks & sds slices
|
|
|
Jerome Quelin <jquelin@mongueurs.net> writes:
| Quote: | are you using soft partitions? since i'm using a two-way mirrors,
this would force me to create the following metadevices (for
example):
==> which is quite clumsy - 5 metadevices for a lone usable
metadevice!
|
Why ? What is the problem with that ?
bash-3.00# metastat -p
d16 -p d3 -o 16777440 -b 1048576
d3 -m d2 d1 1
d2 1 1 c2d0s7
d1 1 1 c1d0s7
d15 -p d3 -o 12583104 -b 4194304
d14 -p d3 -o 8388768 -b 4194304
d13 -p d3 -o 6291584 -b 2097152
d12 -p d3 -o 5242976 -b 1048576
d11 -p d3 -o 4194368 -b 1048576
d10 -p d3 -o 32 -b 4194304
d1 are d2 are the physical devices, d3 the mirror, and dN where N >=
10 are soft partitions, some of them used as zonepath, others for fs
used only inside non global zones.
I use such configuration for few months (~6/, never had problems.
--
http://solaris-fr.org/home/ |
|
| Back to top |
|
|
Rich Teer
*nix forums Guru
Joined: 21 Feb 2005
Posts: 1038
|
| Posted: Tue Feb 14, 2006 5:50 pm Post subject:
Re: zones, disks & sds slices
|
|
|
On Tue, 14 Feb 2006, Jerome Quelin wrote:
| Quote: | with solaris 10 comes the zones facility. in order to segregate zones,
i'd like to create a filesystem of its own for each of the zones' root
filesystem.
|
That's not strictly necessary, although I can see why you'd want to
do it (limiting the amoutn of space used by each zone for /).
| Quote: | current disks are now 73gb+, but still only 7 slices... if one count /,
/var, swap and metadb, this leaves only 3 slices on the disks.
|
Don't separate out /var; that'll save you a slice.
| Quote: | are you using soft partitions? since i'm using a two-way mirrors, this
would force me to create the following metadevices (for example):
d30 -m d100 d110
d100 1 1 d120
d110 1 1 d130
d120 -p c1t0d0s5 -o 1 -b 8389656
d130 -p c1t1d0s5 -o 1 -b 8389656
==> which is quite clumsy - 5 metadevices for a lone usable metadevice!
|
Rather than doing that, you could make the soft partitions from a mirrored
device. So:
d30 -m d100 d110
d100 1 1 c1t0d0s5
d110 1 1 c1t1d0s5
d120 -p d30 -o 1 -b 8389656
--
Rich Teer, SCNA, SCSA, OpenSolaris CAB member
President,
Rite Online Inc.
Voice: +1 (250) 979-1638
URL: http://www.rite-group.com/rich |
|
| Back to top |
|
|
Darren Dunham
*nix forums Guru
Joined: 22 Feb 2005
Posts: 1120
|
| Posted: Tue Feb 14, 2006 5:07 pm Post subject:
Re: zones, disks & sds slices
|
|
|
Jerome Quelin <jquelin@mongueurs.net> wrote:
| Quote: | are you using soft partitions? since i'm using a two-way mirrors, this
would force me to create the following metadevices (for example):
d30 -m d100 d110
d100 1 1 d120
d110 1 1 d130
d120 -p c1t0d0s5 -o 1 -b 8389656
d130 -p c1t1d0s5 -o 1 -b 8389656
==> which is quite clumsy - 5 metadevices for a lone usable
metadevice! |
Looks like you're building the base as a soft partition. It might be
easier to build the top that way instead.
Create one large mirror. Then use SP metadevices as offsets within the
mirror. You'd have 3 metadevices constant, then one more for each zone.
--
Darren Dunham ddunham@taos.com
Senior Technical Consultant TAOS http://www.taos.com/
Got some Dr Pepper? San Francisco, CA bay area
< This line left intentionally blank to confuse you. > |
|
| Back to top |
|
|
Mauricio Tavares
*nix forums addict
Joined: 23 Feb 2005
Posts: 74
|
| Posted: Tue Feb 14, 2006 3:24 pm Post subject:
Re: zones, disks & sds slices
|
|
|
Lion-O wrote:
Let me butt in since I am just about to be making the same decision you
two are talking about above: either make a slice for each zone or group
them all in one single slice and use some other way to keep them from
using up the space. Lion-O, I do like your approach to use one single
slice but I am still not sure how to implement it. Let's use as an
example a zone assigned to be a mail server (postfix and friends and
maildir for mail storage). How would I go about it? I mean, would I then
create a quota for root and postfix and normal users (for their
maildirs) and whoever ever needs to have one?
--
Mauricio raub-kudria-com
(if you need to email me, use this address =) |
|
| Back to top |
|
|
Casper H.S. Dik
*nix forums Guru
Joined: 20 Feb 2005
Posts: 1634
|
| Posted: Tue Feb 14, 2006 3:24 pm Post subject:
Re: zones, disks & sds slices
|
|
|
Lion-O <nosp@m.catslair.org> writes:
| Quote: | Groups and users. And since your root (and any other slice you mount in the
zone) will be picked up as a seperate filesystem...
|
ufs doesn't support group quotas, other may.
| Quote: | So dump *all* zones on a single slice (this way it can't affect your host) and
make sure you properly configure them. When utilizing logadm and the likes the
risk of the system filling itself up are very rare IMO. This goes double when
you're using inherited directories.
|
Or single filesystem in ZFS speak; ZFS will make this all much
easier and flexible. (It allows you to pool all the unused bits whereas
separate slices do not)
Casper
--
Expressed in this posting are my opinions. They are in no way related
to opinions held by my employer, Sun Microsystems.
Statements on Sun products included here are not gospel and may
be fiction rather than truth. |
|
| Back to top |
|
|
Lion-O
*nix forums Guru
Joined: 24 Mar 2005
Posts: 346
|
| Posted: Tue Feb 14, 2006 2:49 pm Post subject:
Re: zones, disks & sds slices
|
|
|
| Quote: | Coming next to my mind would be volume/space control but even that is
something I somewhat fail to see since you can utilize quota's which should
also be usable from within the zone itself.
wrong. you can't use quota for restricting zone space - or can you?
|
I don't see why not.. Check the mountpoints from within the zone, you'll
notice that "/ on / read/write/setuid/devices/intr/largefiles... etc.". Why
wouldn't it be possible to add quota support to it ?
| Quote: | afaik, quota only let you enforce space used by users on a filesystem basis.
|
Groups and users. And since your root (and any other slice you mount in the
zone) will be picked up as a seperate filesystem...
| Quote: | it's quite simple: i don't want the filling of a zone root filesystem filling
the root filesystem of another zone. indeed, a filesystem full for / or /var
on a zone can prevent the zone from working correctly.
|
So dump *all* zones on a single slice (this way it can't affect your host) and
make sure you properly configure them. When utilizing logadm and the likes the
risk of the system filling itself up are very rare IMO. This goes double when
you're using inherited directories.
Next I'd utilize quota's so that your users can't abuse the system to extreme
extends. You could use 1 seperate slice for your data storage, and give each
zone its own space through lofs (indeed, that was a typo  ).
Apart from that I don't see much alternatives.
--
Groetjes, Peter
..\\ PGP/GPG key: http://www.catslair.org/pubkey.asc |
|
| Back to top |
|
|
Jerome Quelin
*nix forums beginner
Joined: 14 Feb 2006
Posts: 3
|
| Posted: Tue Feb 14, 2006 1:49 pm Post subject:
Re: zones, disks & sds slices
|
|
|
On 06/02/14 12:42 -0000, Lion-O wrote:
| Quote: | Coming next to my mind would be volume/space control but even that is something
I somewhat fail to see since you can utilize quota's which should also be
usable from within the zone itself.
|
wrong. you can't use quota for restricting zone space - or can you?
afaik, quota only let you enforce space used by users on a filesystem
basis.
it's quite simple: i don't want the filling of a zone root filesystem
filling the root filesystem of another zone. indeed, a filesystem full
for / or /var on a zone can prevent the zone from working correctly.
| Quote: | [ Seperate slice for zones ]
so i'd like to know how you guys are dealing with this? since zfs is still
not there - and won't be able to be installed on boot disks anyway at first -
i'm stuck with sds...
I wonder what you hope to gain by dumping it all on seperate slices ?
|
i can't use separate slices - there aren't enough for that.
i want separate filesystems for the different zones' root.
and it's not possible with a slice per filesystem, since there's not
a lot of available slices.
hence my question: how do you guys deal with that? using sds soft
partitions? or other? any feedback on the used method?
| Quote: | So for me... I dumped my zones on a place in /opt and utilized lsof to make
them access a specific slice which I reserved for data. Several zones share
this setup, I really fail to see the need to seperate all this. IMO it only
makes your life unecessarily difficult.
|
(i guess you meant lofs)
==> then, nothing prevents a zone from filling up your data slice. and
other zones will suffer from this.
let's take the example of a consolidated server with a zone for
development, and a zone for production. given the licence price of some
softs, it's not that a stupid thing to do. sometimes, we tests scripts
or corner cases or debug instructions that can fill the filesystems,
either with various files or insane logging level.
==> do you really think that i'm making my own life more difficult by
segregating each zone in its own filesystem?
regards,
jérôme
--
jquelin@mongueurs.net |
|
| Back to top |
|
|
Lion-O
*nix forums Guru
Joined: 24 Mar 2005
Posts: 346
|
| Posted: Tue Feb 14, 2006 12:42 pm Post subject:
Re: zones, disks & sds slices
|
|
|
[ Seperate slice for zones ]
| Quote: | so i'd like to know how you guys are dealing with this? since zfs is still
not there - and won't be able to be installed on boot disks anyway at first -
i'm stuck with sds...
|
I wonder what you hope to gain by dumping it all on seperate slices ? You
won't gain much when it comes to security since you'll be unable to mount these
slices with, for example, the 'nodev' option since that will render your
zone(s) useless. Then you have 'noexec' but that will be easily ignored within
the zone itself.
Coming next to my mind would be volume/space control but even that is something
I somewhat fail to see since you can utilize quota's which should also be
usable from within the zone itself.
So for me... I dumped my zones on a place in /opt and utilized lsof to make
them access a specific slice which I reserved for data. Several zones share
this setup, I really fail to see the need to seperate all this. IMO it only
makes your life unecessarily difficult.
--
Groetjes, Peter
..\\ PGP/GPG key: http://www.catslair.org/pubkey.asc |
|
| Back to top |
|
|
Jerome Quelin
*nix forums beginner
Joined: 14 Feb 2006
Posts: 3
|
| Posted: Tue Feb 14, 2006 8:29 am Post subject:
zones, disks & sds slices
|
|
|
hey,
with solaris 10 comes the zones facility. in order to segregate zones,
i'd like to create a filesystem of its own for each of the zones' root
filesystem.
current disks are now 73gb+, but still only 7 slices... if one count /,
/var, swap and metadb, this leaves only 3 slices on the disks.
so i'd like to know how you guys are dealing with this? since zfs is
still not there - and won't be able to be installed on boot disks anyway
at first - i'm stuck with sds...
are you using soft partitions? since i'm using a two-way mirrors, this
would force me to create the following metadevices (for example):
d30 -m d100 d110
d100 1 1 d120
d110 1 1 d130
d120 -p c1t0d0s5 -o 1 -b 8389656
d130 -p c1t1d0s5 -o 1 -b 8389656
==> which is quite clumsy - 5 metadevices for a lone usable metadevice!
so, how are you dealing with this situation?
thanks for your help,
jérôme
--
jquelin@mongueurs.net |
|
| Back to top |
|
|
Google
|
|
| Back to top |
|
|
|
|
The time now is Thu Jan 08, 2009 11:57 am | All times are GMT
|
|
Shops | Read Free Manga Online | Debt Consolidation | Debt Consolidation | Guitar Lessons
|
|
Copyright © 2004-2005 DeniX Solutions SRL
|
|
|
|
Other DeniX Solutions sites:
Unix/Linux blog |
electronics forum |
medicine forum |
science forum |
|
|
Privacy Policy
|
Powered by phpBB © 2001, 2005 phpBB Group
|
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
Forum index
»
*nix
»
Solaris
