We have created apt and yum repositories for nginx users on Linux.
They are available on Index of /packages/
for CentOS/RHEL 5 and 6, Ubuntu 10.04 and Debian 6. There are plans to
update packages as soon
as new stable nginx versions are released.
To add nginx yum repository fnginx.repo file has to be created in
/etc/yum.cond.d/ directory.
In CentOS case nginx.repo should look like:
On Wed, 24 Aug 2011 13:22:58 +0400, Sergey B. wrote:
Hi,
We have created apt and yum repositories for nginx users on Linux.
They are available on Index of /packages/
for CentOS/RHEL 5 and 6, Ubuntu 10.04 and Debian 6. There are plans
to update packages as soon
as new stable nginx versions are released.
Hi,
There are already “official” packages for Debian, and therefore also
Ubuntu…
What is the difference?
There are already “official” packages for Debian, and therefore also Ubuntu…
What is the difference?
All these packages was built with same configure options, with
all standard modules (except modules depend on additional
libraries like geoip or xslt) and without third-party modules.
Each linux distribution has own release cycle, own policy,
so we have several native nginx packages for major
linux distribution built with different options and sometimes
quite old. It is convenient to build last stable version of
nginx with same parameters and with most of standard modules,
first of all from support point of view both for us and, i hope,
for nginx users.
As for debian and ubuntu you need to switch to unstable
or backports repository to get a fresh nginx version and this is not
what most users want, for example, on production.
Nevertheless nobody intend to compete with linux packagers,
these package are not “better” then official ones, they are more
predictable
It is convenient to build last stable version of
nginx with same parameters and with most of standard modules,
Meaning you created yet another repository with packages configured
according to what you thing are the best choices. What are they? Which
modules? What’s after --configure? And can i add other modules at
runtime or disable existing ones?
Otherwise, i still prefer compiling from source, but that’s me.
Nice to have the packages straight from the source though, instead of
through a distro.
On Thu, 25 Aug 2011 12:51:16 +0400, Sergey B. wrote:
[…] It is convenient to build last stable version of
nginx with same parameters and with most of standard modules,
first of all from support point of view both for us and, i hope,
for nginx users.
I’m completely agree !
As for debian and ubuntu you need to switch to unstable
or backports repository to get a fresh nginx version and this is not
what most users want, for example, on production.
Yes, you’re right. This is another debate, but I think switching to
testing repo is good, since testing version is now almost considered as
stable.
It is convenient to build last stable version of
nginx with same parameters and with most of standard modules,
Meaning you created yet another repository with packages configured
according to what you thing are the best choices. What are they? Which
modules? What’s after --configure?
On Wed, 2011-08-24 at 13:22 +0400, Sergey B. wrote:
Hi,
We have created apt and yum repositories for nginx users on Linux. They are
available on Index of /packages/
for CentOS/RHEL 5 and 6, Ubuntu 10.04 and Debian 6.
Is there a particular reason there are both CentOS and RHEL
repositories? To the best of my knowledge, CentOS = RHEL.
Sure, one repository is a symlink to another at present. Two paths
is an insurance against differences in the future, though they are
hardly
probable. And since i am building and testing rpms on centos it is
unfair
to name it rhel only.
[nginx]
[skip]
gpgcheck=0
Maybe it will be better to sign provided RPMs?
rpms are signed already by the key with
fingerprint 573B FD6B 3D8F BC64 1079 A6AB ABF5 BD82 7BD9 BF62
uid nginx signing key [email protected]
you can get public one from keyserver (by gpg --recv-key 7BD9BF62) and
import it
with rpm --import. Without public key import gpgcheck=1 would create
more problems
then it would solve.
One small suggestion for nginx.conf. Rather than using:
include /etc/nginx/conf.d/*;
I’d suggest
include /etc/nginx/conf.d/*.conf;
This prevents backup/temporary files that are typically created by
editors from being processed as valid configuration (e.g. default~, #default#, etc).
Hi!
1.0.6 is still not there, though it is avaible for a day already.
When I tried to build it myself, I’ve got a error:
/usr/bin/install: cannot stat
`/home/boris/rpmbuild/SOURCES/nginx.vh.default’: No such file or
directory
The same problem is with files logrotate (which is renamed to
nginx.logrotate) and nginx.vh.ssl.
Please, check your SRPM correctness.
#default#, etc).
I agree, It makes sense to replace * with *.conf. Added in 1.0.6.
There may be the situation when you have not touched default nginx.conf,
so it would be upgraded, but config files without .conf suffix were
added
to conf.d dir, and they would not be included after upgrade. I have no
idea how
to fix it with post/pre install scripts correctly, so check your configs
after upgrade.
It’d be nice to have symlinks for other flavors of RHEL, such as SL
(Scientyfic Linux) and CL (Cloud Linux). They should work as RHEL or
CentOS, but .list file must be modified.