source: branches/fc13-dev/server/doc/ @ 1673

Last change on this file since 1673 was 1672, checked in by ezyang, 14 years ago
More updates from installing b-m and o-f.
File size: 15.1 KB
1# This document is a how-to for installing a Fedora server.
2# It is semi-vaguely in the form of a shell script, but is not really
3# runnable as it stands.
5set -e -x
7# Some commands should be run as the scripts-build user, not root.
9alias asbuild="sudo -u scripts-build"
11# Old versions of this install document advised setting
12# NSS_NONLOCAL_IGNORE=1 anytime you're setting up anything, e.g. using
13# yum, warning that useradd will query LDAP in a stupid way that makes
14# it hang forever.  As of Fedora 13, this does not seem to be a problem,
15# so it's been removed from the instructions.  If an install is hanging,
16# though, try adding NSS_NONLOCAL_IGNORE.
18# This is actually just "pick an active scripts server".  It can't be
19# because our networking config points that domain
20# at localhost, and if our server is not setup at that point things
21# will break.
24# 'branch' is the current svn branch you are on.  You want to
25# use trunk if your just installing a new server, and branches/fcXX-dev
26# if your preparing a server on a new Fedora release.
29# 'server' is the public hostname of your server, for SCP'ing files
30# to and from.
33# Start with a Scripts kickstarted install of Fedora (install-fedora)
35# Take updates, reboot if there's a kernel update.
37    yum update
39# Check out the svn repository. Configure svn not to cache
40# credentials.
42# Copy over root's dotfiles from one of the other machines.
43# Perhaps a useful change is to remove the default aliases
44    cd /root
45    ls -l .bashrc
46    ls -l .ldapvirc
47    ls -l .screenrc
48    ls -l .ssh
49    ls -l .vimrc
50    ls -l .k5login
51    # Trying to scp from server to server won't work, as scp
52    # will attempt to negotiate a server-to-server connection.
53    # Instead, scp to your trusted machine as a temporary file,
54    # and then push to the other server
55scp -r root@$source_server:~/{.bashrc,.ldapvirc,.screenrc,.ssh,.vimrc,.k5login} .
56scp -r {.bashrc,.ldapvirc,.screenrc,.ssh,.vimrc,.k5login} root@$server:~
58# Install the initial set of credentials (to get Kerberized logins once
59# krb5 is installed).  Otherwise, SCP'ing things in will be annoying.
60#   o You probably installed the machine keytab long ago
61    ls -l /etc/krb5.keytab
62#     Use ktutil to combine the host/ and
63#     host/ keys with host/ in
64#     the keytab.  Do not use 'k5srvutil change' on the combined keytab
65#     or you'll break the other servers. (real servers only).  Be
66#     careful about writing out the keytab: if you write it to an
67#     existing file the keys will just get appended.  The correct
68#     credential list should look like:
69#       ktutil:  l
70#       slot KVNO Principal
71#       ---- ---- ---------------------------------------------------------------------
72#          1    5 host/
73#          2    3 host/
74#          3    2      host/
75#   o Replace the ssh host keys with the ones common to all scripts servers (real servers only)
76    ls -l /etc/ssh/*key*
77#     You can do that with:
78scp root@$source_server:/etc/ssh/*key* .
79scp *key* root@$server:/etc/ssh/
80    service sshd reload
82# Check out the scripts /etc configuration
83    # backslash to make us not use the alias
84    cd /root
85    \cp -a etc /
87# NOTE: You will have just lost DNS resolution and the ability
88# to do password SSH in.  If you managed to botch this step without
89# having named setup, you can do a quick fix by frobbing /etc/resolv.conf
90# with a non address for the DNS server.  Be sure to revert it once
91# you have named.
93# NOTE: You can get password SSH back by editing /etc/ssh/sshd_config (allow
94# password auth) and /etc/pam.d/sshd (comment out the first three auth
95# lines).  However, you should have the Kerberos credentials in place
96# so as soon as you install the full set of Scripts packages, you'll get
97# Kerberized logins.
99# Make sure network is working.  If this is a new server name, you'll
100# need to add it to /etc/hosts and
101# /etc/sysconfig/network-scripts/route-eth1.  Kickstart should have
102# configured eth0 and eth1 correctly; use service network restart
103# to add the new routes in route-eth1.
104    service network restart
105    route
106    ifconfig
107    cat /etc/hosts
108    cat /etc/sysconfig/network-scripts/route-eth1
110# This is the point at which you should start updating scriptsified
111# packages for a new Fedora release.  Consult 'upgrade-tips' for more
112# information.
113    yum install -y scripts-base
114    # Some of these packages are naughty and clobber some of our files
115    cd /etc
116    svn revert resolv.conf hosts sysconfig/openafs
118# Replace rsyslog with syslog-ng by doing:
119    rpm -e --nodeps rsyslog
120    yum install -y syslog-ng
121    chkconfig syslog-ng on
123# Fix the openafs /usr/vice/etc <-> /etc/openafs mapping.
124    echo "/afs:/usr/vice/cache:10000000" > /usr/vice/etc/cacheinfo
125    echo "" > /usr/vice/etc/ThisCell
127# [TEST SERVER] If you're installing a test server, this needs to be
128# much smaller; the max filesize on XVM is 10GB.  Pick something like
129# 500000. Also, some of the AFS parameters are kind of retarded (and if
130# you're low on disk space, will actually exhaust our inodes).  Edit
131# these parameters in /etc/sysconfig/openafs
133# Test that zephyr is working
134    chkconfig zhm on
135    service zhm start
136    echo 'Test!' | zwrite -d -c scripts -i test
138# Install the full list of RPMs that users expect to be on the
139# servers.
140rpm -qa --queryformat "%{Name}.%{Arch}\n" | sort > packages.txt
141# arrange for packages.txt to be passed to the server, then run:
142# --skip-broken will (usually) prevent you from having to sit through
143# several minutes of dependency resolution until it decides that
144# it can't install /one/ package.
145    cat packages.txt | xargs yum install -y --skip-broken
147# Check which packages are installed on your new server that are not
148# in the snapshot, and remove ones that aren't needed for some reason
149# on the new machine.  Otherwise, aside from bloat, you may end up
150# with undesirable things for security, like sendmail.
151    rpm -qa --queryformat "%{Name}.%{Arch}\n" | grep -v kernel | sort > newpackages.txt
152    diff -u packages.txt newpackages.txt | grep -v kernel | less
153    # here's a cute script that removes all extra packages
154    diff -u packages.txt newpackages.txt | grep -v kernel | grep '+' | cut -c2- | grep -v "@" | grep -v "++" | xargs yum erase -y
156# We need an upstream version of cgi which we've packaged ourselves, but
157# it doesn't work with the haskell-platform package which expects
158# explicit versions.  So temporarily rpm -e the package, and then
159# install it again after you install haskell-platform.  [Note: You
160# probably won't need this in Fedora 15 or something, when the Haskell
161# Platform gets updated.]
162    rpm -e ghc-cgi-devel ghc-cgi
163    yum install -y haskell-platform
164    yumdownloader ghc-cgi
165    yumdownloader ghc-cgi-devel
166    rpm -i ghc-cgi*1.8.1*.rpm
168# Check out the scripts /usr/vice/etc configuration
169    cd /root/vice
170    \cp -a etc /usr/vice
172# Install the full list of perl modules that users expect to be on the
173# servers.
174    cd /root
175    export PERL_MM_USE_DEFAULT=1
176    cpan # this is interactive, enter the next two lines
177        o conf prerequisites_policy follow
178        o conf commit
179# on a reference server
180perldoc -u perllocal | grep head2 | cut -f 3 -d '<' | cut -f 1 -d '|' | sort -u | perl -ne 'chomp; print "notest install $_\n" if system("rpm -q --whatprovides \"perl($_)\" >/dev/null 2>/dev/null")' > perl-packages.txt
181# arrange for perl-packages.txt to be transferred to server
182    cat perl-packages.txt | perl -MCPAN -e shell
184# Install the Python eggs and Ruby gems and PEAR/PECL doohickeys that are on
185# the other servers and do not have RPMs.
186# The general mode of operation will be to run the "list" command
187# on both servers, see what the differences are, check if those diffs
188# are packaged up as rpms, and install them (rpm if possible, native otherwise)
189# - Look at /usr/lib/python2.6/site-packages and
190#           /usr/lib64/python2.6/site-packages for Python eggs and modules.
191#   There will be a lot of gunk that was installed from packages;
192#   easy-install.pth in /usr/lib/ will tell you what was easy_installed.
193#   First use 'yum search' to see if the relevant package is now available
194#   as an RPM, and install that if it is.  If not, then use easy_install.
195#   Pass -Z to easy_install to install them unzipped, as some zipped eggs
196#   want to be able to write to ~/.python-eggs.  (Also makes sourcediving
197#   easier.)
198    cat /usr/lib/python2.6/site-packages/easy-install.pth
199# - Look at `gem list` for Ruby gems.
200#   Again, use 'yum search' and prefer RPMs, but failing that, 'gem install'.
201#       ezyang: rspec-rails depends on rspec, and will override the Yum
202#       package, so... don't use that RPM yet
203gem list --no-version > gem.txt
204    gem list --no-version | diff gem.txt - | grep "<" | cut -c3- | xargs gem install
205# - Look at `pear list` for Pear fruits (or whatever they're called).
206#   Yet again, 'yum search' for RPMs before resorting to 'pear install'.  Note
207#   that for things in the beta repo, you'll need 'pear install package-beta'.
208#   (you might get complaints about the php_scripts module; ignore them)
209pear list | tail -n +4 | cut -f 1 -d " " > pear.txt
210    pear config-set preferred_state beta
211    pear channel-update
212    pear list | tail -n +4 | cut -f 1 -d " " | diff pear.txt - | grep "<" | cut -c3- | xargs pear install
213# - Look at `pecl list` for PECL things.  'yum search', and if you must,
214#   'pecl install' needed items. If it doesn't work, try 'pear install
215#   pecl/foo' or 'pecl install foo-beta' or those two combined.
216pecl list | tail -n +4 | cut -f 1 -d " " > pecl.txt
217    pecl list | tail -n +4 | cut -f 1 -d " " | diff pecl.txt - | grep "<" | cut -c3- | xargs pecl install --nodeps
219# Setup some Python config
220    echo 'import site, os.path; site.addsitedir(os.path.expanduser("~/lib/python2.6/site-packages"))' > /usr/lib/python2.6/site-packages/00scripts-home.pth
222# Install the credentials.  There are a lot of things to remember here.
223# Be sure to make sure the permissions match up (ls -l on an existing
224# server!).
225scp root@$source_server:{/etc/{sql-mit-edu.cfg.php,daemon.keytab,pki/tls/private/scripts.key,signup-ldap-pw,whoisd-password},/home/logview/.k5login} .
226scp daemon.keytab signup-ldap-pw whoisd-password sql-mit-edu.cfg.php root@$server:/etc
227scp scripts.key root@$server:/etc/pki/tls/private
228scp .k5login root@$server:/home/logview
229    chown afsagent:afsagent /etc/daemon.keytab
230#   o The daemon.scripts keytab (will be daemon.scripts-test for test)
231    ls -l /etc/daemon.keytab
232#   o The SSL cert private key (real servers only)
233    ls -l /etc/pki/tls/private/scripts.key
234#   o The LDAP password for the signup process (real servers only)
235    ls -l /etc/signup-ldap-pw
236#   o The whoisd password (real servers only)
237    ls -l /etc/whoisd-password
238#   o Make sure logview's .k5login is correct (real servers only)
239    cat /home/logview/.k5login
241# Spin up OpenAFS.  This will fail if there's been a new kernel since
242# when you last tried.  In that case, you can hold on till later to
243# start OpenAFS.  This will take a little bit of time;
244    service openafs-client start
246# Check that fs sysname is correct.  You should see, among others,
247# 'amd64_fedoraX_scripts' (vary X) and 'scripts'. If it's not, you
248# probably did a distro upgrade and should update /etc/sysconfig/openafs.
249    fs sysname
251# [TEST SERVER] If you are setting up a test server, pay attention to
252# /etc/sysconfig/network-scripts and do not bind scripts' IP address.
253# You will also need to modify:
254#   o /etc/ldap.conf
255#       add: host
256#   o /etc/nss-ldapd.conf
257#       replace: uri *****
258#       with: uri ldap://
259#   o /etc/openldap/ldap.conf
260#       add: URI ldap://
261#            BASE dc=scripts,dc=mit,dc=edu
262#   o /etc/httpd/conf.d/vhost_ldap.conf
263#       replace: VhostLDAPUrl ****
264#       with: VhostLDAPUrl "ldap://,dc=scripts,dc=mit,dc=edu"
265#   o /etc/postfix/virtual-alias-{domains,maps}
266#       replace: server_host *****
267#       with: server_host = ldap://
268# to use instead of localhost.
269# XXX: someone should write sed scripts to do this
271# [TEST SERVER] If you are setting up a test server, afsagent's cronjob
272# will attempt to be renewing with the wrong credentials
273# (daemon.scripts). Change this:
274    vim /home/afsagent/renew # replace all mentions of
276# Set up replication (see ./install-ldap).
277# You'll need the LDAP keytab for this server: be sure to chown it
278# fedora-ds after you create the fedora-ds user
279    ls -l /etc/dirsrv/keytab
280    cat install-ldap
282# Make the services dirsrv, nslcd, nscd, postfix, and httpd start at
283# boot. Run chkconfig to make sure the set of services to be run is
284# correct.
285    chkconfig dirsrv on
286    chkconfig nslcd on
287    chkconfig nscd on
288    chkconfig postfix on
289    chkconfig httpd on
291# Check sql user credentials (needs to be done after LDAP is setup)
292    chown sql /etc/sql-mit-edu.cfg.php
294# Postfix doesn't actually deliver mail; fix this
295    cd /etc/postfix
296    postmap virtual
298# Munin might not be monitoring packages that were installed after it
299    munin-node-configure --suggest --shell | sh
301# Run fmtutil-sys --all, which does something that makes TeX work.
302# (Note: this errors on XeTeX which is ok.)
303    fmtutil-sys --all
305# Ensure that PHP isn't broken:
306    mkdir /tmp/sessions
307    chmod 01777 /tmp/sessions
309# Ensure fcgid isn't broken (should be 755)
310    ls -l /var/run/mod_fcgid
312# Fix etc by making sure none of our config files got overwritten
313    cd /etc
314    svn status -q
315    # Some usual candidates for clobbering include nsswitch.conf and
316    # sysconfig/openafs
318# ThisCell got clobbered, replace it with
319    echo "" > /usr/vice/etc/ThisCell
321# Reboot the machine to restore a consistent state, in case you
322# changed anything. (Note: Starting kdump fails (this is ok))
324# [OPTIONAL] Your machine's hostname is baked in at install time;
325# in the rare case you need to change it: it appears to be in:
326#   o /etc/sysconfig/network
327#   o your lvm thingies; probably don't need to edit
329# [TEST SERVER] More stuff for test servers
330#   - You need a self-signed SSL cert.  Generate with:
331    openssl req -new -x509 -keyout /etc/pki/tls/private/scripts.key -out /etc/pki/tls/certs/scripts.cert -nodes
332#     Also make /etc/pki/tls/certs/ca.pem match up
333#   - Make (/etc/aliases) root mail go to /dev/null, so we don't spam people
334#   - Edit /etc/httpd/conf.d/scripts-vhost-names.conf to have
335#     be an accepted vhost name
336#   - Look at the old test server and see what config changes are floating around
338# XXX: our SVN checkout should be updated to use
339# (repository and etc) once serving actually works.
340    cd /etc
341    svn switch --relocate svn://$source_server/ svn://
342    cd /usr/vice/etc
343    svn switch --relocate svn://$source_server/ svn://
344    cd /srv/repository
345    asbuild svn switch --relocate svn://$source_server/ svn://
346    asbuild svn up # verify works
Note: See TracBrowser for help on using the repository browser.