Bug 1295491

Summary:	fusor-installer NTP check may leave system with a broken /etc/resolv.conf
Product:	Red Hat Quickstart Cloud Installer	Reporter:	John Matthews <jmatthew>
Component:	fusor-installer	Assignee:	Jason Montleon <jmontleo>
Status:	CLOSED CURRENTRELEASE	QA Contact:	Tasos Papaioannou <tpapaioa>
Severity:	high	Docs Contact:
Priority:	unspecified
Version:	1.0	CC:	arubin, dgao, jmontleo, tpapaioa
Target Milestone:	TP2	Keywords:	Triaged
Target Release:	1.0
Hardware:	Unspecified
OS:	Unspecified
Whiteboard:
Fixed In Version:		Doc Type:	Bug Fix
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2016-10-03 19:08:41 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:
Bug Depends On:
Bug Blocks:	1291836

Description John Matthews 2016-01-04 16:54:46 UTC

Description of problem:

I ran into a situation where the NTP check in fusor-installer failed and my /etc/resolv.conf only contained entries for the Sat VM, therefore I couldn't resolve other addresses and install couldn't proceed.

It looks like we might be modifying /etc/resolv.conf then doing the NTP check, if it fails exiting with a busted /etc/resolv.conf

Version-Release number of selected component (if applicable):

rpm -qa | grep fusor-installer
fusor-installer-0.0.14-23.el7.noarch

How reproducible:

Steps to Reproduce:
1. New VM
2. Run fusor-installer
3. Enter bad info for NTP server
4. Agree to stop fusor-installer
5. Re-run fusor-installer
6. Enter a valid NTP host
7. See how it complains about NTP host can't be reached
8. Agree to stop fusor-installer

Look at /etc/resolv.conf

# cat /etc/resolv.conf
; generated by /usr/sbin/dhclient-script
nameserver 192.168.52.10
nameserver 192.168.52.10

Issue is that the first run changes /etc/resolv.conf to:
nameserver 192.168.52.10
nameserver 192.168.121.1

2nd run changes to:
nameserver 192.168.52.10
nameserver 192.168.52.10

The removal of 192.168.121.1 for this case is what's breaking DNS resolution.

Actual results:

Example of /etc/resolv.conf prior to running fusor-installer:

cat /etc/resolv.conf
; generated by /usr/sbin/dhclient-script
nameserver 192.168.121.1
search localhost

Run fusor-installer for first time

1. Proceed with the values shown
2. Network interface | eth1
3. IP address | 192.168.52.10
4. Hostname | sat61fusorjwm.example.com
5. Network mask | 255.255.255.0
6. Network address | 192.168.52.0
7. Host Gateway | 192.168.121.1
8. DHCP range start | 192.168.52.11
9. DHCP range end | 192.168.52.254
10. DHCP Gateway | 192.168.52.1
11. DNS forwarder | 192.168.121.1
12. Domain | example.com
13. Foreman URL | https://sat61fusorjwm.example.com
14. NTP sync host | foo.example.com
15. Timezone | UTC
16. BMC feature enabled | false
17. BMC default provider | ipmitool
18. Do not configure networking
19. Do not configure firewall
20. Cancel Installation
1
Starting networking setup
Networking setup has finished
WARNING!! - NTP sync host "foo.example.com" does not appear to be valid!
Do you want to continue anyway? [Yes/No]
no
Exiting installation!
Something went wrong! Check the log for ERROR-level output
The full log is at /var/log/katello-installer/katello-installer.log

real 0m42.417s
user 0m5.936s
sys 0m2.223s

# cat /etc/resolv.conf
; generated by /usr/sbin/dhclient-script
nameserver 192.168.52.10
nameserver 192.168.121.1

(DNS Resolution still works after first run)

Re-run fusor-installer
1. Proceed with the values shown
2. Network interface | eth1
3. IP address | 192.168.52.10
4. Hostname | sat61fusorjwm.example.com
5. Network mask | 255.255.255.0
6. Network address | 192.168.52.0
7. Host Gateway | 192.168.121.1
8. DHCP range start | 192.168.52.11
9. DHCP range end | 192.168.52.254
10. DHCP Gateway | 192.168.52.10
11. DNS forwarder | 192.168.52.10
12. Domain | example.com
13. Foreman URL | https://sat61fusorjwm.example.com
14. NTP sync host | 0.rhel.pool.ntp.org
15. Timezone | UTC
16. BMC feature enabled | false
17. BMC default provider | ipmitool
18. Do not configure networking
19. Do not configure firewall
20. Cancel Installation
1
Starting networking setup
Networking setup has finished
WARNING!! - NTP sync host "0.rhel.pool.ntp.org" does not appear to be valid!
Do you want to continue anyway? [Yes/No]
no
Exiting installation!
Something went wrong! Check the log for ERROR-level output
The full log is at /var/log/katello-installer/katello-installer.log

# cat /etc/resolv.conf
; generated by /usr/sbin/dhclient-script
nameserver 192.168.52.10
nameserver 192.168.52.10

DNS resolution is now broken since we removed the entry for the external resolver.

Comment 1 John Matthews 2016-01-04 17:04:49 UTC

In terms of fixing this issue, I think we need to look at the code for adding the Satellite IP to /etc/resolv.conf.

We need the puppet code to be idempotent.  I'm guessing issue is around that area, likely not related to actual NTP check.

Comment 3 Jason Montleon 2016-01-18 17:30:38 UTC

https://github.com/fusor/fusor-installer/pull/41/files

Comment 4 John Matthews 2016-01-22 20:19:47 UTC

Compose of 1/22/16

Comment 5 Tasos Papaioannou 2016-01-25 17:17:19 UTC

Description and PR # are identical to https://bugzilla.redhat.com/show_bug.cgi?id=1295489 , which is VERIFIED on RHCI-6.0-RHEL-7-20160118.t.1. Marking this one as verified too.