Re: Linux

On Thu, Nov 4, 2010 at 12:18 PM, Marco Colombo <pgsql(at)esiway(dot)net> wrote:
> On 11/04/2010 04:00 PM, Michael Gould wrote:
>>
>> I know that this is probably a "religion" issue but we are looking to
>> move Postgres to a Linux server. We currently have a Windows 2008 R2
>> active directory and all of the other servers are virtualized via VMWare
>> ESXi. One of the reasons is that we want to use a 64 bit Postgres server
>> and the UUID processing contrib module does not provide a 64 bit version
>> for Windows. I would also assume that the database when properly tuned
>> will probably run faster in a *inx environment.
>>
>> What and why should I look at certain distributions? It appears from
>> what I read, Ubanta is a good desktop but not a server.
>>
>> Best Regards
>>
>
> Just find one that ships with the latest PG, to save you some work. Unless
> you plan to compile & install PG manually, in that case, any major
> distribution would do. For production use, how long your version will be
> supported for (security updates) is likely to be the most important item in
> your checklist. I use CentOS.

Note that if you'll be running in a mixed server environment, and you
want to use slony replication, it's a good idea to just build pgsql
and slony from source. For instance on Ubuntu (and i'd assume all
debian systems) the pg_config is always from the latest pg version
supported by that distro. slony can't properly build on those
machines against anything but the latest release. Also, it allows you
to make sure of things like int dates are on all machines, etc. Where
I work we have older db servers running Centos and newer ones running
Ubuntu, and the only way to get slony and pg 8.3 happy there was
building from source. Luckily with pgsql it's a freaking snap to have
a configure.local with all the switches for slony and postgresql ready
to go.