0. Plan time in infrastructure(a)lists.fedoraprojects.org
1. Open ticket on infrastructure for downtime.
Updates will occur during day
Reboots will occur during evening
2. Send email to devel-announce, announce, infrastructure
3. Update servers during working hours and work out issues in ticket.
** releng updates the following boxes:
cvs01, pkgs01, nfs01, bnfs01, bxen*,
x86-*, ppc*, koji*, db03, xb-01,
compose-*, sign-vault01
4. Change DNS to turn off proxy on bodhost01 (or similar external
proxy server).
5. Reboot bodhost01
6. Confirm proxy is working on bodhost/fix issues.
7. Change proxy dns to only go to bodhost01
8. Turn off nagios for servers.
9. Turn off nagios-external for services.
10. Reboot order counts
11. releng deals with the boxes listed above unless told otherwise.
12. reboots with database servers first
xen15: db02
xen12: db01
13. reboot PHX2 boxes
xen03:
xen04:
xen06:
xen07:
xen09:
xen10:
xen11:
xen13:
backup01:
14. reboot Outside boxes (can be in parallel to PHX2)
cnode01:
cnode02:
cnode03:
ibiblio01:
internetx01:
osuosl01:
people01:
serverbeach1:
serverbeach2:
serverbeach3:
serverbeach4:
serverbeach5:
telia1:
tummy1:
15. reboot
bastion.fedoraproject.org
log into bastion1 from outside system
log into bastion2 from outside world
log into xen05 from bastion01
bastion01:
sudo su /usr/sbin/puppetd --disable
sudo su /sbin/service openvpn start
bastion02
sudo su /sbin/service openvpn start
xen05
sudo /sbin/shutdown -r now
once xen05/bastion2 server is back up, we can
bastion01:
sudo su /sbin/service openvpn stop
sudo su /usr/sbin/puppetd --enable
16. reboot puppet01
log into bastion2 from outside world
ssh xen14
sudo /sbin/shutdown -r now
17. re-enable DNS for proxy servers
test proxy servers from puppet01
edit dns in git puppet
make ns1
18. re-enable nagios on internal/external
19. Setup transifex agent on app servers: app01 app02 app03 app04 app07
sudo -u transifex /var/lib/transifex/ssh-add.sh -f
20. Log and report problems to list.
21. Close ticket.
--
Stephen J Smoogen.
“The core skill of innovators is error recovery, not failure avoidance.”
Randy Nelson, President of Pixar University.
"We have a strategic plan. It's called doing things.""
— Herb Kelleher, founder Southwest Airlines