Partitioning

[mbrijun]

Hi everyone,

The server that I am going to install Zimbra (open source) has an internal RAID1 array of 18Gb and and external RAID10 array of 80Gb. What would be the recommended partitioning schema? Thank you.

Martin

[Kristaps]

RAID1

/BOOT - 100 MB
/SWAP - 2x RAM
/ - ALL REST

RAID10
/OPT - ALL

[Kristaps]

From your post it seems that you did not read the documentation are you not good with Linux.

1. Download Opens SuSE 10.2
2. Install text mode – Fallow this The Perfect Setup - OpenSuSE 10.2 (32-bit) | HowtoForge - Linux Howtos and Tutorials documentation till page 5 chapter - 3 Install Some Software (stop before 3 Install Some Software)do not forget to set partitions to
RAID1
/BOOT - 100 MB
/SWAP - 2x RAM (2 times moor then ram)
/ - ALL REST

RAID10
/OPT – ALL
3. Execute: shutdown -r now
4. Execute: yast2 -i findutils readline libgcc glibc-devel findutils-locate gcc flex lynx compat-readline4 db-devel wget gcc-c++ make mc xntp perl-HTML-Parser perl-Net-DNS perl-Digest-SHA1 fetchmail cURL libidn GMP compat-libstdc++
5. Execute: shutdown -r now
6. Continue with: http://www.zimbra.com/forums/install...ete-guide.html (STEP 1 IS ALREADY DONE)

[Dirk]

Looks about right to me.

[mbrijun]

Thanks very much for the answers. It makes my life easy, as I do not want to deal with RAID10 during the initial install, but rather create it afterwards.

[dijichi2]

personally I would split root partition up a bit further - /var should always be on it's own for instance.

possibly to use the 80gb for /opt/zimbra/store instead? that way you get to use the extra IO bw of the rootdisk for amavis/sql stuff and use full raid10 bw for just mailstore. I don't know if this is good idea, just a suggestion.

[Kristaps]

1. Download Opens SuSE 10.2
2. Install text mode – Fallow this The Perfect Setup - OpenSuSE 10.2 (32-bit) | HowtoForge - Linux Howtos and Tutorials documentation till page 5 chapter - 3 Install Some Software (stop before 3 Install Some Software)do not forget to set partitions to
RAID1
/BOOT - 75 MB
/SWAP - 2x RAM (2 times moor then ram)
/ - ALL REST

/VAR - Depends from OS and usage (/ - ALL REST will be much easier for beginner)

RAID10
/opt/zimbra/store – ALL
3. Execute: shutdown -r now
4. Execute: yast2 -i findutils readline libgcc glibc-devel findutils-locate gcc flex lynx compat-readline4 db-devel wget gcc-c++ make mc xntp perl-HTML-Parser perl-Net-DNS perl-Digest-SHA1 fetchmail cURL libidn GMP compat-libstdc++
5. Execute: shutdown -r now
6. Continue with: http://www.zimbra.com/forums/install...ete-guide.html (STEP 1 IS ALREADY DONE)

Extra IO is always good

Originally Posted by dijichi2

personally I would split root partition up a bit further - /var should always be on it's own for instance.

possibly to use the 80gb for /opt/zimbra/store instead? that way you get to use the extra IO bw of the rootdisk for amavis/sql stuff and use full raid10 bw for just mailstore. I don't know if this is good idea, just a suggestion.

[dijichi2]

/VAR - Depends from OS and usage (/ - ALL REST will be much easier for beginner)

Aside from the obvious danger of a beginner installing a core corporate communications system (!), /var is quite important to separate. It is the most volatile fs area, and the one most likely to impact server stability. The truly paranoid further separate out /var/log and /var/adm.

[Kristaps]

Ok,

Most probably you are more experience with those thinks so what should look like configuration of Zimbra Server FS in case when you have:

RAID1 – 20GB ~
RAID10 – 80 GB ~

In my case I have very similar configuration and I would like to know the perfect configuration and maybe explanations for beginners! – in *NIX environment.

[dijichi2]

hey, I'm no authority on the subject, just giving my opinion

your original layout is a sensible easy desktop layout. for a server that is running important software I feel it's worth the extra effort of thinking about your disk layout. seperating /usr these days is rarely useful, especially under linux where it is almost always local, but seperating /var is a no-brainer. for example, if not maintained, on a busy server /var/log can easily grow to fill the disk and if not in a seperate partition can bring the system to it's knees.

running lvm, veritas or some other volume management is also worth investigating - this means you can resize your partitions dynamically at a later date if needs be.

ps

3. Execute: shutdown -r now
4. Execute: yast2 -i findutils readline libgcc glibc-devel findutils-locate gcc flex lynx compat-readline4 db-devel wget gcc-c++ make mc xntp perl-HTML-Parser perl-Net-DNS perl-Digest-SHA1 fetchmail cURL libidn GMP compat-libstdc++
5. Execute: shutdown -r now

why the two reboots? I can't see anything there that would necessitate a reboot. i've lost track of what OS does what, but 'init 6' is the safer way to reboot a server.