7945
Comment:
|
25180
|
Deletions are marked like this. | Additions are marked like this. |
Line 1: | Line 1: |
These steps are listed in approximately the order in which they should be performed; please try to maintain that. === Set Up Out Of Band Access === All machines owned by hcoop should, if possible, have some out-of-band mechanism for: |
#pragma section-numbers off These steps are listed in approximately the order in which they should be performed; please try to maintain that as you add to it. <<TableOfContents>> = List the Machine on the Wiki = The hostname of the machine should be decided through a Members Poll (accessible from members portal) such as [[https://members.hcoop.net/portal/poll?id=31]]. Add the machine to the [[Hardware]] page. It is a very good idea to photograph the front and back panels of the machine and put those images on the wiki page; that way remote admins and people in the data center can be sure they're talking about the same ports. Add the machine to the [[IpAddresses]] page. = Set Up Out Of Band Access = All machines owned by hcoop should, if possible, have some [[http://en.wikipedia.org/wiki/Out-of-band_infrastructure|out-of-band mechanism]] for: |
Line 15: | Line 30: |
=== Add a DNS entry for the server === Straightforward. === Install Debian === We use Debian. Install it. === Compile a Kernel === It is generally a good idea for hcoop to compile its own kernels. Regarding statically-compiled kernels, see StaticallyCompiledKernels for some opinions. === Install the AFS Client === First, give our preferences to {{{debconf}}}: {{{ debconf openafs-client/thiscell HCOOP.NET debconf openafs-client/dynroot true debconf openafs-client/cachesize 500000 # cache size in kB; default is way too small }}} You should install the {{{module-assistant}}}, {{{build-essential}}}, {{{module-init-tools}}}, {{{openafs-client}}}, {{{openafs-krb5}}}, {{{openafs-modules-source}}}, {{{openafs-dbg}}}, {{{openafs-doc}}}, {{{libopenafs-dev}}}, packages from {{{/afs/hcoop.net/common/debian/}}}. Here is a block of commands to cut and paste if you are lazy: {{{ apt-get install krb5-user libkrb5-dev module-assistant build-essential module-init-tools mkdir -p /tmp/openafs-packages cd /tmp/openafs-packages scp ssh.hcoop.net:/afs/hcoop.net/common/debian/openafs/1.4.6/\*.deb ./ dpkg -i \ |
If there's _anything_ server-specific, please add an entry under "Specific Machines" on page AdminArea and document what it is. Rebooting procedures are an ideal candidate for this. = Add a DNS entry for the server = This is done as follows: 1. Have our DomTool2 CVS repository from SourceForge checked out (for this you need a SourceForge account of course, and write access to the "hcoop" project on SourceForge) 1. Edit domtool2/lib/hcoop.dtl and add definition for "HOSTNAME_ip" (search for "deleuze_ip" and just copy the line to new name) 1. TODO: How to recompile and install new domtool with HOSTNAME_ip defined 1. Edit ''/afs/hcoop.net/user/h/hc/hcoop/.domtool/hcoop.net'' to add the new DNS entry, using HOSTNAME_ip (again, can use deleuze_ip as example) 1. To apply DomTool configuration, run '''DOMTOOL_USER=hcoop domtool hcoop.net''' in the ~hcoop/.domtool/ directory = Install Debian = We use Debian GNU. Here are the installation notes to help you: 1. Find Debian stable image (whichever is 'stable' at time of installation) 1. Prepare a USB stick to boot from (can do it manually or with convenient tool called "unetbootin") 1. In system BIOS, choose 'Auto-power on on power restore' (if there is such option), and see if you can make USB stick to not be the first disk (when it's the first disk, it gets assigned device name /dev/sda and makes the installation a tiny bit harder) 1. See which network card is in the server, if it requires non-free firmware, the package needs to be manually copied from Debian's non-free repository onto the install media (example is package "firmware-bnx2" for Broadcom NetXtremeII cards (http://packages.debian.org/sid/all/firmware-bnx2/download)). Once package is on the media, the install procedure will, if it is needed, automatically find and install it 1. For timezone, use timezone where the server is physically located, and answer Yes to "Is the hardware clock set to GMT?" 1. Choose manual network configuration, specifying the choosen hostname, IP and network details as listed on the IpAddresses page 1. Partition disks. Most often, this comes town to creating identical partitions on all disks that are part of RAID1, and creating RAID arrays as inthe following example (can probably reuse it verbatim in your scenario): {{{ Example: 2x 160 GB system disks System disk 1: sda1: primary, beginning, 1 GB, ext3, /boot sda2: primary, beginning, 8 GB, use as phys. volume for RAID (swap space: 1 GB x number of proc. cores) sda3: primary, beginning, all available space, use as phys. volume for RAID System disk 2: sdb1: primary, beginning, 1 GB, ext3, unmounted sdb2: primary, beginning, 8 GB, use as phys. volume for RAID (swap, same size as above) sdb3: primary, beginning, all available space, use as phys. volume for RAID Then, after RAID partitions have been assigned, new option "Configure RAID" will appear at the top of the partitioning menu. We add the two devices in RAID 1 mode: md0: sda2 and sdb2 md1: sda3 and sdb3 Then, they appear in the partitions list and are configured as follows: md0: swap md1: ext3, / }}} * As seen, /boot partition is not on RAID. This is intentional as /boot on RAID is problematic. But sda1 (/boot) will be synced onto sdb1 (by cping the files and running grub-install on it) periodically via cron, so that sdb1 can be used for booting too, in case sda fails. (TODO: Cron job/Puppet recipe to copy that does this) * Users & password setup: set root password and choose "No" at "Create regular user account?" prompt. If the installation does not let you continue without creating a regular user, create "root2" with the same password as root. The password should not be an official password, but a strong temporary string. (TODO: Puppet recipe that manages passwords and admin's SSH keys). * If /dev/sda is the USB stick and not the first disk, do not install GRUB to the Master Boot Record of /dev/sda. Instead, answer No at the prompt and choose /dev/sdb as the device. Then, take USB stick out, edit /boot/grub/menu.lst to replace references to hd(1,0) with hd(0,0), run '''update-grub.conf''' and '''grub-install /dev/sdb'''. No other tunings (to /etc/fstab or mdadm.conf) are needed as, if you used the partitioning example, no direct partitions occur in fstab, and for mdadm -- it uses UUIDs instead of partition names anyway. * In tasksel, at the end of installation, do not select any package category, not even "Standard system" = Booting into the new machine = When the machine boots for the first time, run: {{{ dpkg-reconfigure debconf # (choose interface: Dialog, priority: Low). apt-get install less sudo vim etckeeper changetrack lm-sensors openssh-server debsums logcheck bzip2 denyhosts }}} Verify that disks performance is as expected using '''sync; sync; hdparm -tT /dev/sdX'''. Activate etckeeper as documented on EtcKeeper. Edit ''/etc/default/changetrack'' and set '''AUTO_TRACK_ALL_CONFFILES=yes'''. Edit ''/etc/tripwire/twcfg.txt'' and set '''MAILNOVIOLATIONS =false'''. Initialize the database with '''tripwire --init'''. (If tripwire is installed) Edit ''/etc/aliases'' and set "root" alias to "logs@hcoop.net", and possibly other addresses, separated by commas. (logs@ is an aliasMulti, defined in ~hcoop/.domtool/hcoop.net and lists people who want to receive verbose system logs). Run '''sensors-detect''' to see if the kernel has appropriate thermal modules for the server, and add any drivers detected to ''/etc/modules''. For all ext partitions, run '''tune2fs -j -c0 -i0 /dev/sdXX''' (and /dev/mdX for RAID arrays). == Tune the /etc/apt/sources.list == {{{ cat > /etc/apt/sources.list <<\EOF deb http://mirror.peer1.net/debian/ lenny main deb-src http://mirror.peer1.net/debian/ lenny main deb http://security.debian.org/ lenny/updates main deb-src http://security.debian.org/ lenny/updates main deb http://volatile.debian.org/debian-volatile lenny/volatile main deb-src http://volatile.debian.org/debian-volatile lenny/volatile main EOF apt-get update apt-get dist-upgrade }}} == Remove lame directories == {{{ sudo rm /cdrom sudo rm /media/cdrom sudo rm /media/floppy sudo rmdir /media/cdrom[0-9] sudo rmdir /media/floppy[0-9] sudo rmdir /media sudo rmdir /opt }}} = Compile a Kernel (not needed) = Here's an example for kernel 2.6.31.9, adjust accordingly: {{{ apt-get install make gcc patch bin86 kernel-package libncurses5-dev fakeroot cd /usr/local/src wget http://grsecurity.net/stable/grsecurity-2.1.14-2.6.31.9-200912191011.patch wget http://www.kernel.org/pub/linux/kernel/v2.6/linux-2.6.31.9.tar.bz2 tar jxf linux-2.6.31.9.tar.bz2 patch -p0 < grsecurity-2.1.14-2.6.31.9-200912191011.patch cd linux-2.6.31.9 cp /some/existing/config .config make oldconfig make menuconfig # (if any manual tuning needed) time CONCURRENCY_LEVEL=8 fakeroot make-kpkg --initrd kernel_image >& ../build.log }}} = Install the AFS Client = The AFS client gets very unhappy if the partition holding {{{/var/cache/openafs}}} fills up. To ensure that this can't happen, we'll create a 2GB file and mount it there using the loopback device. This gives the openafs client a partition-in-a-file all to itself that no other process can interfere with. First, create the file: {{{ dd if=/dev/zero of=/var/cache/openafs.ext3 bs=1M count=2K chmod go-rwx /var/cache/openafs.ext3 mke2fs -F /var/cache/openafs.ext3 tune2fs -j -i0 -c0 /var/cache/openafs.ext3 }}} Then mount it. Note: we could mount it directly on {{{/var/cache/openafs}}}, but if we did that and for some reason it failed to mount, the openafs client would just write files into that directory anyways. We want to know immediately if the mount fails, so we'll make {{{/var/cache/openafs}}} a symlink to a subdirectory of the new partition. {{{ mkdir /var/cache/openafs.mnt echo -e '/var/cache/openafs.ext3\t/var/cache/openafs.mnt\text3\tloop\t1\t1' >> /etc/fstab mount /var/cache/openafs.mnt/ mkdir -p /var/cache/openafs.mnt/cache/ rm -rf /var/cache/openafs ln -s /var/cache/openafs.mnt/cache /var/cache/openafs }}} Then, give our preferences to {{{debconf}}}: {{{ debconf-set-selections <<\EOF openafs-client openafs-client/thiscell string hcoop.net openafs-client openafs-client/thiscell seen true openafs-client openafs-client/dynroot boolean true openafs-client openafs-client/dynroot seen true openafs-client openafs-client/cachesize string 500000 openafs-client openafs-client/cachesize seen true openafs-client openafs-client/cell-info string openafs-client openafs-client/cell-info seen true openafs-client openafs-client/run-client boolean true openafs-client openafs-client/run-client seen true EOF }}} You should install the {{{module-assistant}}}, {{{build-essential}}}, {{{module-init-tools}}}, {{{openafs-client}}}, {{{openafs-krb5}}}, {{{openafs-modules-source}}}, {{{openafs-doc}}}, {{{libopenafs-dev}}}, and {{{kstart}}} packages. Here is a block of commands to cut and paste if you are lazy: {{{ apt-get install krb5-user libkrb5-dev module-init-tools kstart sudo \ module-assistant build-essential bison flex debhelper mkdir -p /tmp/openafs-packages cd /tmp/openafs-packages scp ssh.hcoop.net:/afs/hcoop.net/common/debian/openafs/1.4.6/\*.deb ./ dpkg -i \ |
Line 48: | Line 213: |
openafs-dbg*.deb \ | |
Line 51: | Line 215: |
cd /tmp rm -rf /tmp/openafs-packages |
cd /tmp rm -rf /tmp/openafs-packages |
Line 66: | Line 230: |
/etc/init.d/module-init-tools start | depmod /etc/init.d/module-init-tools start |
Line 77: | Line 242: |
=== Install Packages === | = Install Packages = |
Line 82: | Line 247: |
dpkg -i /afs/hcoop.net/common/debian/libnss-ptdb/*.deb dpkg -i /afs/hcoop.net/common/debian/libpam-afs-session/*.deb dpkg -i /afs/hcoop.net/common/debian/libpam-krb5/*.deb dpkg -i /afs/hcoop.net/common/debian/fsr/*.deb }}} The first three packages are explained below; the last one is the {{{fsr}}} command (recursive "{{{fs}}}"). === Configure Kerberos === You should copy {{{/etc/krb5.conf}}} from deleuze to the new server. This is VERY IMPORTANT. What is NOT in this file is also almost as important as what IS in this file, so think three times before adding or removing anything. === Configure Name Service === |
dpkg -i /afs/hcoop.net/user/m/me/megacz/public/libnss-afs/libnss-afs*.deb dpkg -i /afs/hcoop.net/common/debian/libpam-afs-session/*.deb dpkg -i /afs/hcoop.net/common/debian/libpam-krb5/*.deb dpkg -i /afs/megacz.com/debian/fsr*.deb dpkg -i /afs/megacz.com/debian/krb5-user/{krb5-user,libk}*.deb }}} The first three packages are explained below; the fourth one is the {{{fsr}}} command (recursive "{{{fs}}}"). The last line installs a fixed version of kadmin which understands DNS entries. = Install Network Time Protocol Daemon = Kerberos and AFS will not work correctly unless the clocks of the client and server are synchronized to within a certain tolerance. Therefore, it is important for us to have a daemon running that keeps the clock set properly. '''This step is not optional'''. {{{ apt-get install ntp }}} = Install LDAP Support = Logins etc. will not work correctly unless libpam-ldap is installed and configured: {{{ apt-get install libpam-ldap }}} Debconf answers: {{{ debconf-set-selections <<\EOF libpam-ldap shared/ldapns/base-dn string dc=hcoop,dc=net libpam-ldap shared/ldapns/ldap-server string ldap://69.90.123.67/ libpam-ldap libpam-ldap/pam_password select exop libpam-ldap libpam-ldap/rootbinddn string cn=admin,dc=hcoop,dc=net libpam-ldap libpam-ldap/dbrootlogin boolean true libpam-ldap libpam-ldap/override boolean true libpam-ldap shared/ldapns/ldap_version select 3 libpam-ldap libpam-ldap/dblogin boolean false EOF }}} You will also need to know LDAP admin password; see /etc/pam_ldap.secret on one of existing servers and re-type the password into the password prompt. = Configure Kerberos = '''''VERY IMPORTANT''''': put exactly the following in {{{/etc/krb5.conf}}} -- no more, no less {{{ [libdefaults] default_realm = HCOOP.NET kdc_timesync = 1 forwardable = true proxiable = true rdns = no # undocumented option to disable reverse DNS lookups [logging] default = FILE:/proc/self/fd/2 }}} We distribute our Kerberos configuration via DNS, so it is very important that we do not "hardwire" the settings on any of the servers (except the KDCs themselves). If we did, we wouldn't notice at first, but strange problems would crop up as soon as the DNS settings were changed. So, it is important that we put only the bare minimum amount of information in {{{krb5.conf}}}. = Configure Name Service = |
Line 98: | Line 312: |
1. the userid for a given username 2. the username for a userid |
1. the userid for a given username and vice versa 2. the groupid for a given groupname and vice versa |
Line 104: | Line 318: |
The {{{libnss-ptdb}}} package lets linux use the AFS user database (the {{{ptserver}}} or protection server) as a name service. To make {{{ptserver}}} our primary choice for name service, edit {{{/etc/nsswitch.conf}}} and change the following three lines to look like this: {{{ passwd: ptdb files group: afspag files shadow: files }}} === Configure PAM === |
The {{{libnss-afs}}} package lets linux use the AFS user database (the {{{ptserver}}} or protection server) as a name service and makes PAGs show up as a special group. To enable these changes, edit {{{/etc/nsswitch.conf}}} and change the {{{passwd}}} and {{{group}}} lines to look like this: {{{ passwd: afs files group: afs files shadow: files }}} = Install Name Service Caching Daemon = It is highly recommended to install {{{nscd}}} in order to get good performance out of {{{libnss-afs}}}. {{{ apt-get install nscd }}} Unfortunately there is a grevious bug in the DNS caching mechanism in etch's nscd (see [[http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=467609|this]]); so we must disable it until it is fixed. To do this, edit {{{/etc/nscd.conf}}} and change the line {{{ sed -i 's_enable-cache.*hosts.*yes_enable-cache hosts no_' /etc/nscd.conf }}} We prefer to run nscd as a runit service so that it does not go down (except on deleuze, where it must be started strictly after AFS in the boot sequence). {{{ apt-get install runit mkdir /var/service/nscd cat <<EOF > /var/service/nscd/run #!/bin/sh exec nscd -d EOF mkdir /var/service/nscd/log cat <<EOF > /var/service/nscd/log/run #!/bin/bash svlogd -tt /var/log/nscd/ EOF mkdir /var/log/nscd chmod +x /var/service/nscd/log/run chmod +x /var/service/nscd/run dpkg-divert --rename /etc/init.d/nscd ln -s /usr/bin/sv /etc/init.d/nscd }}} = Configure PAM = |
Line 120: | Line 372: |
FIXME Mostly this consists of copying mire's {{{/etc/pam.d/*}}}, although it would be a good idea to state precisely which parts of that need to be copied. Also should document the mechanism used on deleuze and other non-member servers to restrict logins to admins only. === Configure SSH === |
Here's the usual PAM setup: /etc/pam.d/common-account: {{{ account sufficient pam_unix.so account required pam_ldap.so account required pam_krb5.so debug # temporary line for emergencies #account required pam_unix.so account required pam_access.so }}} /etc/pam.d/common-auth: {{{ auth sufficient pam_krb5.so debug forwardable ignore_root auth optional pam_afs_session.so program=/usr/bin/aklog debug auth required pam_unix.so nullok_secure try_first_pass # temporary line for emergencies #auth required pam_unix.so nullok_secure auth required pam_env.so }}} /etc/pam.d/common-password: {{{ password sufficient pam_krb5.so password required pam_unix.so nullok obscure min=4 max=8 md5 shadow try_first_pass }}} /etc/pam.d/common-session: {{{ session requisite pam_limits.so session required pam_unix_session.so # Unix module just logs access session optional pam_krb5.so session optional pam_afs_session.so program=/usr/bin/aklog debug }}} /etc/pam.d/login (Add to beginning of file): {{{ auth required pam_listfile.so item=user sense=allow file=/etc/login.restrict onerr=succeed }}} /etc/pam.d/ssh (Add just before {{{@include common-auth}}} line): {{{ # sshd does not consult the "auth" section of pam when # GssapiAuthentication=yes, even if UsePAM=yes. Therefore, we add the # check to the "account" section as well. account requisite pam_listfile.so item=user sense=allow file=/etc/login.restrict onerr=succeed auth requisite pam_listfile.so item=user sense=allow file=/etc/login.restrict onerr=succeed }}} If the machine is intended for user logins, DO NOT create /etc/login.restrict. If the machine is only intended for admin logins, then create the file /etc/login.restrict with the following contents: {{{ adamc_admin docelic_admin megacz_admin mwolson_admin ntk_admin }}} = Configure SSH = == Configure SSH Client == Insert these lines in {{{/etc/ssh/ssh_config}}} so that ''outbound'' ssh connections will always try to use Kerberos if available: {{{ Host * GSSAPIAuthentication yes GSSAPIDelegateCredentials no }}} == Configure SSH Server == |
Line 135: | Line 467: |
rm -f /etc/krb5.keytab # just in case kadmin -r HCOOP.NET ank -randkey host/server.hcoop.net@HCOOP.NET ktadd -k /etc/krb5.keytab host/server.hcoop.net@HCOOP.NET quit |
REALM=HCOOP.NET ADMIN=myself_admin # your admin username SERVER=server.hcoop.net rm -f /etc/krb5.keytab # important -- if it already exists the new key will merely be appended kadmin -p $ADMIN@$REALM -r $REALM -q "ank -randkey host/$SERVER@$REALM" kadmin -p $ADMIN@$REALM -r $REALM -q "ktadd -k /etc/krb5.keytab host/$SERVER@$REALM" |
Line 144: | Line 477: |
Then these lines to {{{/etc/ssh/sshd_config}}}: | Then add these lines to the bottom of {{{/etc/ssh/sshd_config}}}: |
Line 149: | Line 482: |
GSSAPICleanupCredentials no UsePAM yes }}} === Optional Steps === ==== Performance-Tune the OpenAFS Client ==== |
GSSAPICleanupCredentials yes }}} Finally, restart the ssh server: {{{ /etc/init.d/ssh restart }}} = Populate sudoers = Don't forget to give all of the admins lines in {{{/etc/sudoers}}}. Each line should look like: {{{ user_admin ALL=(ALL) NOPASSWD: ALL }}} = Set Up Some Cron Scripts = /etc/cron.daily/hcoop-clean-tmp: {{{ #!/bin/sh # # Clean /tmp periodically. # # Edit $TMPTIME in /etc/default/rcS to change the maximal age of /tmp entries # before they are removed. exec /afs/hcoop.net/common/etc/scripts/hcoop-clean-tmp }}} = Optional Steps = == Install commonly-used packages == {{{ apt-get install \ xbase-clients # provides xauth, without which "ssh -Y" will not work dpkg-dev-el # provide debian-changelog-mode }}} == Performance-Tune the OpenAFS Client == |
Line 159: | Line 528: |
==== runit ==== | == runit == |
Line 165: | Line 534: |
4. Runit captures the daemon's {{{stdout}}} and either sends it to a logger (if specified) or else displays it in the process name (output of {{ps}}) | 4. Runit captures the daemon's {{{stdout}}} and either sends it to a logger (if specified) or else displays it in the process name (output of {{{ps}}}) |
Line 178: | Line 547: |
==== dnscache ==== You can install the dnscache package to make the server self-sufficient for dns resolution purposes (it acts as a tiny dns server just for localhost). This improves the reliability of the overall infrastructure. There is a copy of this package in {{{/afs/megacz.com/debian/dnscache/}}}; the author of the software recently changed its license, so it will be a standard package in the next release of debian (it may even be in etch-backports already; when it is, this paragraph should be updated to recommend that instead). |
== dnscache == You can install the dnscache package to make the server self-sufficient for dns resolution purposes (it acts as a tiny dns server just for localhost). This improves the reliability of the overall infrastructure. |
Line 183: | Line 552: |
Here are the instructions for configuring it. Make sure that bind9 (if running) is only listening to {{{127.0.0.1}}} and the public IP address of the machine. We tell dnscache to listen on {{{127.0.0.2}}} so as to avoid conflicts with bind. {{{ apt-get install djbdns # If needed: addgroup --system Gdnscache adduser --system Gdnscache --ingroup Gdnscache # Create /etc/service/dnscache dnscache-conf Gdnscache Gdnscache /etc/service/dnscache 127.0.0.2 # Change default listen address 127.0.0.1 to .2 perl -pi -e 's/\.1/.2/' /etc/service/dnscache/env/IP # Let dnscache answer queries only from 127.0.0.2 mv /var/dnscache/root/ip/127.0.0.1 /var/dnscache/root/ip/127.0.0.2 sv restart dnscache }}} Then modify {{{/etc/resolv.conf}}}, replacing the {{{nameserver}}} lines with: {{{ nameserver 127.0.0.2 }}} == /etc/hosts == If not present already: {{{ echo '127.0.0.1 localhost' > /etc/hosts }}} == ssmtp == Life is simpler when you run {{{ssmtp}}}. You can direct the mail stream either to {{{deleuze}}} (preferred) or to a copy of {{{exim}}} running locally (but why bother running it?). Be sure to enable {{{FromLineOverride}}}, which ships defaulted to "off" in Debian. {{{ apt-get install ssmtp sed -i 's_FromLineOverride.*_FromLineOverride=YES_' /etc/ssmtp/ssmtp.conf }}} == noatime == By default, Linux will write to the disk in order to update the atime ("access time") every time a file is ''read from''; this substantially degrades performance. You can disable this behavior by editing {{{/etc/fstab}}} {{{ # <file system> <mount point> <type> <options> <dump> <pass> /dev/hda1 / ext3 defaults,noatime,errors=remount-ro 0 1 }}} This is especially important on filesystems which are used to store AFS volumes. == locales == If you installed debian via {{{debootstrap}}}, you will be missing the {{{locales}}} package and your locale will not be set. You can fix this with: {{{ debconf-set-selections <<\EOF locales locales/default_environment_locale select en_US locales locales/default_environment_locale seen true locales locales/locales_to_be_generated multiselect en_US ISO-8859-1 locales locales/locales_to_be_generated multiselect seen true EOF apt-get install locales }}} == etckeeper == {{{ apt-get install etckeeper cd /etc etckeeper init etckeeper commit "Initial checkin" git gc }}} == nitpicks == 1. Debian's installer seems to want to put an entry for the machine's own hostname in /etc/hosts, resolving to 127.0.0.1. You'll probably want to remove it. |
These steps are listed in approximately the order in which they should be performed; please try to maintain that as you add to it.
Contents
- List the Machine on the Wiki
- Set Up Out Of Band Access
- Add a DNS entry for the server
- Install Debian
- Booting into the new machine
- Compile a Kernel (not needed)
- Install the AFS Client
- Install Packages
- Install Network Time Protocol Daemon
- Install LDAP Support
- Configure Kerberos
- Configure Name Service
- Install Name Service Caching Daemon
- Configure PAM
- Configure SSH
- Populate sudoers
- Set Up Some Cron Scripts
- Optional Steps
List the Machine on the Wiki
The hostname of the machine should be decided through a Members Poll (accessible from members portal) such as https://members.hcoop.net/portal/poll?id=31.
Add the machine to the Hardware page.
It is a very good idea to photograph the front and back panels of the machine and put those images on the wiki page; that way remote admins and people in the data center can be sure they're talking about the same ports.
Add the machine to the IpAddresses page.
Set Up Out Of Band Access
All machines owned by hcoop should, if possible, have some out-of-band mechanism for:
- Keyboard access
- Screen access
- Power-cycling
Functions 1+2 are typically provided by kvm.hcoop.net (see KvmAccess); assuming you plan on going with that, you should connect the server's keyboard and video to the kvm switch.
Each server has its own solution for 3, usually in the form of a "service processor". You should investigate and document the appropriate service processor settings. If the service processor requires its own IP address, you should name it foo-sp.hcoop.net where foo.hcoop.net is the name of the server.
If there's _anything_ server-specific, please add an entry under "Specific Machines" on page AdminArea and document what it is. Rebooting procedures are an ideal candidate for this.
Add a DNS entry for the server
This is done as follows:
Have our DomTool2 CVS repository from SourceForge checked out (for this you need a SourceForge account of course, and write access to the "hcoop" project on SourceForge)
- Edit domtool2/lib/hcoop.dtl and add definition for "HOSTNAME_ip" (search for "deleuze_ip" and just copy the line to new name)
- TODO: How to recompile and install new domtool with HOSTNAME_ip defined
Edit /afs/hcoop.net/user/h/hc/hcoop/.domtool/hcoop.net to add the new DNS entry, using HOSTNAME_ip (again, can use deleuze_ip as example)
To apply DomTool configuration, run DOMTOOL_USER=hcoop domtool hcoop.net in the ~hcoop/.domtool/ directory
Install Debian
We use Debian GNU.
Here are the installation notes to help you:
- Find Debian stable image (whichever is 'stable' at time of installation)
- Prepare a USB stick to boot from (can do it manually or with convenient tool called "unetbootin")
- In system BIOS, choose 'Auto-power on on power restore' (if there is such option), and see if you can make USB stick to not be the first disk (when it's the first disk, it gets assigned device name /dev/sda and makes the installation a tiny bit harder)
See which network card is in the server, if it requires non-free firmware, the package needs to be manually copied from Debian's non-free repository onto the install media (example is package "firmware-bnx2" for Broadcom NetXtremeII cards (http://packages.debian.org/sid/all/firmware-bnx2/download)). Once package is on the media, the install procedure will, if it is needed, automatically find and install it
- For timezone, use timezone where the server is physically located, and answer Yes to "Is the hardware clock set to GMT?"
Choose manual network configuration, specifying the choosen hostname, IP and network details as listed on the IpAddresses page
Partition disks. Most often, this comes town to creating identical partitions on all disks that are part of RAID1, and creating RAID arrays as inthe following example (can probably reuse it verbatim in your scenario):
Example: 2x 160 GB system disks System disk 1: sda1: primary, beginning, 1 GB, ext3, /boot sda2: primary, beginning, 8 GB, use as phys. volume for RAID (swap space: 1 GB x number of proc. cores) sda3: primary, beginning, all available space, use as phys. volume for RAID System disk 2: sdb1: primary, beginning, 1 GB, ext3, unmounted sdb2: primary, beginning, 8 GB, use as phys. volume for RAID (swap, same size as above) sdb3: primary, beginning, all available space, use as phys. volume for RAID Then, after RAID partitions have been assigned, new option "Configure RAID" will appear at the top of the partitioning menu. We add the two devices in RAID 1 mode: md0: sda2 and sdb2 md1: sda3 and sdb3 Then, they appear in the partitions list and are configured as follows: md0: swap md1: ext3, /
- As seen, /boot partition is not on RAID. This is intentional as /boot on RAID is problematic. But sda1 (/boot) will be synced onto sdb1 (by cping the files and running grub-install on it) periodically via cron, so that sdb1 can be used for booting too, in case sda fails. (TODO: Cron job/Puppet recipe to copy that does this)
Users & password setup: set root password and choose "No" at "Create regular user account?" prompt. If the installation does not let you continue without creating a regular user, create "root2" with the same password as root. The password should not be an official password, but a strong temporary string. (TODO: Puppet recipe that manages passwords and admin's SSH keys).
If /dev/sda is the USB stick and not the first disk, do not install GRUB to the Master Boot Record of /dev/sda. Instead, answer No at the prompt and choose /dev/sdb as the device. Then, take USB stick out, edit /boot/grub/menu.lst to replace references to hd(1,0) with hd(0,0), run update-grub.conf and grub-install /dev/sdb. No other tunings (to /etc/fstab or mdadm.conf) are needed as, if you used the partitioning example, no direct partitions occur in fstab, and for mdadm -- it uses UUIDs instead of partition names anyway.
- In tasksel, at the end of installation, do not select any package category, not even "Standard system"
Booting into the new machine
When the machine boots for the first time, run:
dpkg-reconfigure debconf # (choose interface: Dialog, priority: Low). apt-get install less sudo vim etckeeper changetrack lm-sensors openssh-server debsums logcheck bzip2 denyhosts
Verify that disks performance is as expected using sync; sync; hdparm -tT /dev/sdX.
Activate etckeeper as documented on EtcKeeper.
Edit /etc/default/changetrack and set AUTO_TRACK_ALL_CONFFILES=yes.
Edit /etc/tripwire/twcfg.txt and set MAILNOVIOLATIONS =false. Initialize the database with tripwire --init. (If tripwire is installed)
Edit /etc/aliases and set "root" alias to "logs@hcoop.net", and possibly other addresses, separated by commas. (logs@ is an aliasMulti, defined in ~hcoop/.domtool/hcoop.net and lists people who want to receive verbose system logs).
Run sensors-detect to see if the kernel has appropriate thermal modules for the server, and add any drivers detected to /etc/modules.
For all ext partitions, run tune2fs -j -c0 -i0 /dev/sdXX (and /dev/mdX for RAID arrays).
Tune the /etc/apt/sources.list
cat > /etc/apt/sources.list <<\EOF deb http://mirror.peer1.net/debian/ lenny main deb-src http://mirror.peer1.net/debian/ lenny main deb http://security.debian.org/ lenny/updates main deb-src http://security.debian.org/ lenny/updates main deb http://volatile.debian.org/debian-volatile lenny/volatile main deb-src http://volatile.debian.org/debian-volatile lenny/volatile main EOF apt-get update apt-get dist-upgrade
Remove lame directories
sudo rm /cdrom sudo rm /media/cdrom sudo rm /media/floppy sudo rmdir /media/cdrom[0-9] sudo rmdir /media/floppy[0-9] sudo rmdir /media sudo rmdir /opt
Compile a Kernel (not needed)
Here's an example for kernel 2.6.31.9, adjust accordingly:
apt-get install make gcc patch bin86 kernel-package libncurses5-dev fakeroot cd /usr/local/src wget http://grsecurity.net/stable/grsecurity-2.1.14-2.6.31.9-200912191011.patch wget http://www.kernel.org/pub/linux/kernel/v2.6/linux-2.6.31.9.tar.bz2 tar jxf linux-2.6.31.9.tar.bz2 patch -p0 < grsecurity-2.1.14-2.6.31.9-200912191011.patch cd linux-2.6.31.9 cp /some/existing/config .config make oldconfig make menuconfig # (if any manual tuning needed) time CONCURRENCY_LEVEL=8 fakeroot make-kpkg --initrd kernel_image >& ../build.log
Install the AFS Client
The AFS client gets very unhappy if the partition holding /var/cache/openafs fills up. To ensure that this can't happen, we'll create a 2GB file and mount it there using the loopback device. This gives the openafs client a partition-in-a-file all to itself that no other process can interfere with.
First, create the file:
dd if=/dev/zero of=/var/cache/openafs.ext3 bs=1M count=2K chmod go-rwx /var/cache/openafs.ext3 mke2fs -F /var/cache/openafs.ext3 tune2fs -j -i0 -c0 /var/cache/openafs.ext3
Then mount it. Note: we could mount it directly on /var/cache/openafs, but if we did that and for some reason it failed to mount, the openafs client would just write files into that directory anyways. We want to know immediately if the mount fails, so we'll make /var/cache/openafs a symlink to a subdirectory of the new partition.
mkdir /var/cache/openafs.mnt echo -e '/var/cache/openafs.ext3\t/var/cache/openafs.mnt\text3\tloop\t1\t1' >> /etc/fstab mount /var/cache/openafs.mnt/ mkdir -p /var/cache/openafs.mnt/cache/ rm -rf /var/cache/openafs ln -s /var/cache/openafs.mnt/cache /var/cache/openafs
Then, give our preferences to debconf:
debconf-set-selections <<\EOF openafs-client openafs-client/thiscell string hcoop.net openafs-client openafs-client/thiscell seen true openafs-client openafs-client/dynroot boolean true openafs-client openafs-client/dynroot seen true openafs-client openafs-client/cachesize string 500000 openafs-client openafs-client/cachesize seen true openafs-client openafs-client/cell-info string openafs-client openafs-client/cell-info seen true openafs-client openafs-client/run-client boolean true openafs-client openafs-client/run-client seen true EOF
You should install the module-assistant, build-essential, module-init-tools, openafs-client, openafs-krb5, openafs-modules-source, openafs-doc, libopenafs-dev, and kstart packages. Here is a block of commands to cut and paste if you are lazy:
apt-get install krb5-user libkrb5-dev module-init-tools kstart sudo \ module-assistant build-essential bison flex debhelper mkdir -p /tmp/openafs-packages cd /tmp/openafs-packages scp ssh.hcoop.net:/afs/hcoop.net/common/debian/openafs/1.4.6/\*.deb ./ dpkg -i \ openafs-client*.deb \ openafs-krb5*.deb \ openafs-modules-source*.deb \ openafs-doc*.deb \ libopenafs-dev*.deb cd /tmp rm -rf /tmp/openafs-packages
Once these packages are installed, you will want to run
module-assistant a-i -t openafs-modules
... assuming you compiled your own kernel and the compiled kernel tree resides in /usr/src/linux. If this is not the case, you are on your own.
If the command above completes, it will have created and installed a .deb containing the kernel module. You may need to run
depmod /etc/init.d/module-init-tools start
to refresh whatever module wonkery linux maintains in obscure locations. Once this is figured out (if all else fails, reboot) you should be able to
/etc/init.d/openafs-client start
Do this and check that /afs shows up.
Install Packages
Now that afs is up, you can easily install packages. The block of commands below installs the set of packages which must be on every hcoop server (this list will be expanded as necessary).
dpkg -i /afs/hcoop.net/user/m/me/megacz/public/libnss-afs/libnss-afs*.deb dpkg -i /afs/hcoop.net/common/debian/libpam-afs-session/*.deb dpkg -i /afs/hcoop.net/common/debian/libpam-krb5/*.deb dpkg -i /afs/megacz.com/debian/fsr*.deb dpkg -i /afs/megacz.com/debian/krb5-user/{krb5-user,libk}*.deb
The first three packages are explained below; the fourth one is the fsr command (recursive "fs"). The last line installs a fixed version of kadmin which understands DNS entries.
Install Network Time Protocol Daemon
Kerberos and AFS will not work correctly unless the clocks of the client and server are synchronized to within a certain tolerance. Therefore, it is important for us to have a daemon running that keeps the clock set properly. This step is not optional.
apt-get install ntp
Install LDAP Support
Logins etc. will not work correctly unless libpam-ldap is installed and configured:
apt-get install libpam-ldap
Debconf answers:
debconf-set-selections <<\EOF libpam-ldap shared/ldapns/base-dn string dc=hcoop,dc=net libpam-ldap shared/ldapns/ldap-server string ldap://69.90.123.67/ libpam-ldap libpam-ldap/pam_password select exop libpam-ldap libpam-ldap/rootbinddn string cn=admin,dc=hcoop,dc=net libpam-ldap libpam-ldap/dbrootlogin boolean true libpam-ldap libpam-ldap/override boolean true libpam-ldap shared/ldapns/ldap_version select 3 libpam-ldap libpam-ldap/dblogin boolean false EOF
You will also need to know LDAP admin password; see /etc/pam_ldap.secret on one of existing servers and re-type the password into the password prompt.
Configure Kerberos
VERY IMPORTANT: put exactly the following in /etc/krb5.conf -- no more, no less
[libdefaults] default_realm = HCOOP.NET kdc_timesync = 1 forwardable = true proxiable = true rdns = no # undocumented option to disable reverse DNS lookups [logging] default = FILE:/proc/self/fd/2
We distribute our Kerberos configuration via DNS, so it is very important that we do not "hardwire" the settings on any of the servers (except the KDCs themselves). If we did, we wouldn't notice at first, but strange problems would crop up as soon as the DNS settings were changed. So, it is important that we put only the bare minimum amount of information in krb5.conf.
Configure Name Service
A "name service" is Linux's mechanism for answering these queries:
- the userid for a given username and vice versa
- the groupid for a given groupname and vice versa
- the home directory for a user
- the shell for a user
- what groups a user is in
The libnss-afs package lets linux use the AFS user database (the ptserver or protection server) as a name service and makes PAGs show up as a special group. To enable these changes, edit /etc/nsswitch.conf and change the passwd and group lines to look like this:
passwd: afs files group: afs files shadow: files
Install Name Service Caching Daemon
It is highly recommended to install nscd in order to get good performance out of libnss-afs.
apt-get install nscd
Unfortunately there is a grevious bug in the DNS caching mechanism in etch's nscd (see this); so we must disable it until it is fixed. To do this, edit /etc/nscd.conf and change the line
sed -i 's_enable-cache.*hosts.*yes_enable-cache hosts no_' /etc/nscd.conf
We prefer to run nscd as a runit service so that it does not go down (except on deleuze, where it must be started strictly after AFS in the boot sequence).
apt-get install runit mkdir /var/service/nscd cat <<EOF > /var/service/nscd/run #!/bin/sh exec nscd -d EOF mkdir /var/service/nscd/log cat <<EOF > /var/service/nscd/log/run #!/bin/bash svlogd -tt /var/log/nscd/ EOF mkdir /var/log/nscd chmod +x /var/service/nscd/log/run chmod +x /var/service/nscd/run dpkg-divert --rename /etc/init.d/nscd ln -s /usr/bin/sv /etc/init.d/nscd
Configure PAM
PAM is Linux's mechanism to do the following:
- decide if somebody is who they say they are (authentication; in our case via kerberos)
set up sessions (in the case of AFS, this means creating PAGs)
- change passwords (in our case, changing the password in the KDC)
Here's the usual PAM setup:
/etc/pam.d/common-account:
account sufficient pam_unix.so account required pam_ldap.so account required pam_krb5.so debug # temporary line for emergencies #account required pam_unix.so account required pam_access.so
/etc/pam.d/common-auth:
auth sufficient pam_krb5.so debug forwardable ignore_root auth optional pam_afs_session.so program=/usr/bin/aklog debug auth required pam_unix.so nullok_secure try_first_pass # temporary line for emergencies #auth required pam_unix.so nullok_secure auth required pam_env.so
/etc/pam.d/common-password:
password sufficient pam_krb5.so password required pam_unix.so nullok obscure min=4 max=8 md5 shadow try_first_pass
/etc/pam.d/common-session:
session requisite pam_limits.so session required pam_unix_session.so # Unix module just logs access session optional pam_krb5.so session optional pam_afs_session.so program=/usr/bin/aklog debug
/etc/pam.d/login (Add to beginning of file):
auth required pam_listfile.so item=user sense=allow file=/etc/login.restrict onerr=succeed
/etc/pam.d/ssh (Add just before @include common-auth line):
# sshd does not consult the "auth" section of pam when # GssapiAuthentication=yes, even if UsePAM=yes. Therefore, we add the # check to the "account" section as well. account requisite pam_listfile.so item=user sense=allow file=/etc/login.restrict onerr=succeed auth requisite pam_listfile.so item=user sense=allow file=/etc/login.restrict onerr=succeed
If the machine is intended for user logins, DO NOT create /etc/login.restrict. If the machine is only intended for admin logins, then create the file /etc/login.restrict with the following contents:
adamc_admin docelic_admin megacz_admin mwolson_admin ntk_admin
Configure SSH
Configure SSH Client
Insert these lines in /etc/ssh/ssh_config so that outbound ssh connections will always try to use Kerberos if available:
Host * GSSAPIAuthentication yes GSSAPIDelegateCredentials no
Configure SSH Server
You will need to create a "host principal" for the new server; if you are setting up server.hcoop.net, then it must have the name
host/server.hcoop.net@HCOOP.NET
Add this principal to the KDC like this (execute these commands on the new server, as root, while holding admin tickets):
REALM=HCOOP.NET ADMIN=myself_admin # your admin username SERVER=server.hcoop.net rm -f /etc/krb5.keytab # important -- if it already exists the new key will merely be appended kadmin -p $ADMIN@$REALM -r $REALM -q "ank -randkey host/$SERVER@$REALM" kadmin -p $ADMIN@$REALM -r $REALM -q "ktadd -k /etc/krb5.keytab host/$SERVER@$REALM" chown root:root /etc/krb5.keytab chmod go-rwx /etc/krb5.keytab
Then add these lines to the bottom of /etc/ssh/sshd_config:
GssapiKeyExchange yes GssapiAuthentication yes GSSAPICleanupCredentials yes
Finally, restart the ssh server:
/etc/init.d/ssh restart
Populate sudoers
Don't forget to give all of the admins lines in /etc/sudoers. Each line should look like:
user_admin ALL=(ALL) NOPASSWD: ALL
Set Up Some Cron Scripts
/etc/cron.daily/hcoop-clean-tmp:
# # Clean /tmp periodically. # # Edit $TMPTIME in /etc/default/rcS to change the maximal age of /tmp entries # before they are removed. exec /afs/hcoop.net/common/etc/scripts/hcoop-clean-tmp
Optional Steps
Install commonly-used packages
apt-get install \ xbase-clients # provides xauth, without which "ssh -Y" will not work dpkg-dev-el # provide debian-changelog-mode
Performance-Tune the OpenAFS Client
FIXME: AdamM needs to fill this in
runit
The runit package is a mechanism for starting, stopping, and monitoring daemons. It is an alternative to the traditional /etc/init.d and start-stop-daemon scheme. Its chief advantages are:
It launches daemons with clean process state; the daemon inherits nothing from the administrator invoking the start/stop command because the daemon is not forked as a child of the administrator's shell (rather, a request is sent runit daemon asking it to fork the daemon). This is very important when dealing with tokens and pags.
- Runit monitors the processes that it forks, and restarts them if they die.
- Runit eliminates the need for pidfiles and the associated risk of starting multiple copies of a daemon.
Runit captures the daemon's stdout and either sends it to a logger (if specified) or else displays it in the process name (output of ps)
apt-get install runit
When you move a process from /etc/init.d/ control to runit supervision, you should inform debian that you have done so:
# assuming /var/service/$SERVICE/run is the runit script dpkg-divert --rename /etc/init.d/$SERVICE ln -s /usr/bin/sv /etc/init.d/$SERVICE
This will cause invocations of /etc/init.d/script {start|stop} to do "the right thing".
dnscache
You can install the dnscache package to make the server self-sufficient for dns resolution purposes (it acts as a tiny dns server just for localhost). This improves the reliability of the overall infrastructure.
Starting dnscache via runit is often a good idea; this ensures that it starts early in the boot process and that it is restarted if it dies for any reason.
Here are the instructions for configuring it. Make sure that bind9 (if running) is only listening to 127.0.0.1 and the public IP address of the machine. We tell dnscache to listen on 127.0.0.2 so as to avoid conflicts with bind.
apt-get install djbdns # If needed: addgroup --system Gdnscache adduser --system Gdnscache --ingroup Gdnscache # Create /etc/service/dnscache dnscache-conf Gdnscache Gdnscache /etc/service/dnscache 127.0.0.2 # Change default listen address 127.0.0.1 to .2 perl -pi -e 's/\.1/.2/' /etc/service/dnscache/env/IP # Let dnscache answer queries only from 127.0.0.2 mv /var/dnscache/root/ip/127.0.0.1 /var/dnscache/root/ip/127.0.0.2 sv restart dnscache
Then modify /etc/resolv.conf, replacing the nameserver lines with:
nameserver 127.0.0.2
/etc/hosts
If not present already:
echo '127.0.0.1 localhost' > /etc/hosts
ssmtp
Life is simpler when you run ssmtp. You can direct the mail stream either to deleuze (preferred) or to a copy of exim running locally (but why bother running it?).
Be sure to enable FromLineOverride, which ships defaulted to "off" in Debian.
apt-get install ssmtp sed -i 's_FromLineOverride.*_FromLineOverride=YES_' /etc/ssmtp/ssmtp.conf
noatime
By default, Linux will write to the disk in order to update the atime ("access time") every time a file is read from; this substantially degrades performance. You can disable this behavior by editing /etc/fstab
# <file system> <mount point> <type> <options> <dump> <pass> /dev/hda1 / ext3 defaults,noatime,errors=remount-ro 0 1
This is especially important on filesystems which are used to store AFS volumes.
locales
If you installed debian via debootstrap, you will be missing the locales package and your locale will not be set. You can fix this with:
debconf-set-selections <<\EOF locales locales/default_environment_locale select en_US locales locales/default_environment_locale seen true locales locales/locales_to_be_generated multiselect en_US ISO-8859-1 locales locales/locales_to_be_generated multiselect seen true EOF apt-get install locales
etckeeper
apt-get install etckeeper cd /etc etckeeper init etckeeper commit "Initial checkin" git gc
nitpicks
- Debian's installer seems to want to put an entry for the machine's own hostname in /etc/hosts, resolving to 127.0.0.1. You'll probably want to remove it.