I am encountering a problem while updating my nginx server.
I first configured and built the new version 1.1.11.
I then stopped the server, checked it has been done, executed the make
install and checked the configuration with the ‘-t’ option which was OK.
I then started the new version, but I am still seeing the 1.0.9 number
in
the header of the pages loaded in my browser. I hit F5 several times to
be
sure it wasn’t the cache.
I deleted the whole nginx directory, reinstalled the new version… and
the
headers still got stuck on 1.0.9…
Then… I discovered that the executable used was /usr/local/sbin/nginx
and
not /usr/local/nginx/sbin/nginx… Is the first file a link? Yes, but
not
symbolic…
I wiped it out, recreated it as symbolic and now I get the 1.1.11 up and
running :o)
Sharl, I used the ‘sudo make install’ command to be sure the install
would
get the admin rights.
‘which nginx’ won’t work, but ‘/usr/local/nginx/sbin/nginx -v’ gives me
the
1.1.11 version.
Antonio, I tried your commands, but I already had my browser configured
so
that its cache is deleted when I exit.
I tried again this morning after I booted my computer again and I still
gets the 1.0.9 header version…
I made a chown on the /usr/local/nginx’ directory so the nginx user and
a
group not related to it are the owners of all the files inside. I
remember
having seen some subdirectories (such as ‘conf’) owned by the user root
and
not nginx.
Does that could have an impact? I doubt it, since the ‘other’ rights are
still set to ‘read’. It was maybe one of my ‘security’ tweaks of which I
don’t really know the useness.
If anyone has an idea apart from wiping out the whole directory (which
would be far away the ‘hot update’ feature idea planted in nginx), I
would
be glad to take it…