Commit b75ac057 authored by Robert Ricci's avatar Robert Ricci

Cleanup, and more notes about local-only steps.

parent bc582d4f
......@@ -232,6 +232,7 @@ inetd - In FreeBSD, you need to prevent inetd from rate-limiting connections
environment). Put the following in /etc/rc.conf:
inetd_flags="-wW -R 0"
Local Only:
tftp - Should have the following line in /etc/inetd.conf
tftp dgram udp wait nobody /usr/libexec/tftpd tftpd /tftpboot /proj
(make sure to HUP inetd)
......@@ -245,17 +246,19 @@ cvsupd - Minor changes to images can be distributed at boot time with cvsup.
See doc/newimage.txt for an overview of setting up a sup tree. Make sure to
copy over the old one (if it exists), and make sure cvsupd is running (there's
an example rc.d script in the rc.d/ directory of the testbed CVS tree.) Create
a group named 'root', with any gid. This is because cvsup uses the game of
a group named 'root', with any gid. This is because cvsup uses the name of
the group, rather than its gid, to determine what group the file should belong
to. Since Linux uses 'root' instead of BSD's 'wheel', this is needed for the
Linux sup tree.
Local Only:
syslogd - Normally, sylogd on FreeBSD is run with the '-s' flag to prevent
logging to it over the network. We use network logging, so we need this
feature. Re-enable it by putting:
syslogd_flags=""
in /etc/rc.conf
Local Only:
dhcpd - Need to install the dhcpd config file. The old (deprecated) location was
/usr/site/bin/dhcp/dhcpd.conf. The new location (and the place you should
install it if you used the 'isc-dhcpd' port) is /usr/local/etc/dhcpd.conf .
......@@ -265,6 +268,7 @@ the dhcpd directory of the CVS repository, to generate the dhcpd.conf file.
##### Step 7 - Misc. Files and Services
Local Only:
SNMP MIBs - MIBs go in /usr/local/share/snmp/mibs. In addition to the ones
installed by the ucd-snmp package, you'll need MIBs for Cisco and Intel
switches. You can grab the Cisco MIBs from:
......@@ -280,11 +284,7 @@ to 666. Now, do an snmpwalk of some device (eg. 'snmpwalk cisco1 public') -
this will force the .index file to get rebuilt. Suggestions of better ways to
rebuild this file are welcome!
SSH - If possible, grab the old machine's SSH host keys (from
/etc/ssh/ssh_host*) and HUP sshd. Also, get the root identity and known_hosts
files from the old machine (/root/.ssh/{identity,identity.pub,known_hosts}) -
Make sure to preserve file and directory permissions. You'll probably also want
to add
SSH - You'll probably also want to add
PermitRootLogin yes
to /etc/ssh/sshd_config (and HUP sshd) so that you can log in as root remotely
......@@ -304,7 +304,7 @@ to /etc/ssh/sshd_config (and HUP sshd) so that you can log in as root remotely
All of these logs should be created before you HUP syslogd or reboot - All of
them can be world-readable
cron JOBS: We currently have two cron jobs running for the testbed. Both can be
cron jobs: We currently have two cron jobs running for the testbed. Both can be
run out of /etc/crontab
45 1 * * * root /usr/testbed/sbin/backup
*/5 * * * * root /usr/testbed/sbin/node_status
......@@ -345,6 +345,8 @@ and trees you'll want to make sure to copy over:
/etc/group
/usr/testbed/sup/
/usr/site/
/root/.ssh
/etc/ssh
Right before bringing the new boss node online (if copying from an old boss
node), make sure to have copy over the latest versions of:
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment