Bug 1021435 - sssd does not work properly in pure IPv6 environment
sssd does not work properly in pure IPv6 environment
Status: ASSIGNED
Product: Red Hat Enterprise Linux 7
Classification: Red Hat
Component: sssd (Show other bugs)
7.0
Unspecified Unspecified
medium Severity medium
: rc
: ---
Assigned To: SSSD Maintainers
Namita Soman
:
Depends On:
Blocks:
  Show dependency treegraph
 
Reported: 2013-10-21 05:41 EDT by Patrik Kis
Modified: 2017-06-16 08:29 EDT (History)
10 users (show)

See Also:
Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of:
Environment:
Last Closed:
Type: Bug
Regression: ---
Mount Type: ---
Documentation: ---
CRM:
Verified Versions:
Category: ---
oVirt Team: ---
RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: ---


Attachments (Terms of Use)
sssd log files (3.73 KB, application/x-gzip)
2013-10-21 05:42 EDT, Patrik Kis
no flags Details

  None (edit)
Comment 1 Patrik Kis 2013-10-21 05:42:00 EDT
Created attachment 814506 [details]
sssd log files
Comment 2 Jakub Hrozek 2013-10-21 08:31:51 EDT
This is what I see in the logs:

(Mon Oct 21 05:27:20 2013) [sssd[be[ad.baseos.qe]]] [resolv_gethostbyname_dns_query] (0x0100): Trying to resolve A record of 'sec-ad1.ad.baseos.qe' in DNS
(Mon Oct 21 05:27:20 2013) [sssd[be[ad.baseos.qe]]] [schedule_request_timeout] (0x2000): Scheduling a timeout of 6 seconds
(Mon Oct 21 05:27:20 2013) [sssd[be[ad.baseos.qe]]] [schedule_timeout_watcher] (0x2000): Scheduling DNS timeout watcher
(Mon Oct 21 05:27:20 2013) [sssd[be[ad.baseos.qe]]] [unschedule_timeout_watcher] (0x4000): Unscheduling DNS timeout watcher
(Mon Oct 21 05:27:20 2013) [sssd[be[ad.baseos.qe]]] [resolv_gethostbyname_dns_parse] (0x1000): Parsing an A reply
(Mon Oct 21 05:27:20 2013) [sssd[be[ad.baseos.qe]]] [request_watch_destructor] (0x0400): Deleting request watch
(Mon Oct 21 05:27:20 2013) [sssd[be[ad.baseos.qe]]] [sdap_connect_host_resolv_done] (0x0400): Connecting to ldap://sec-ad1.ad.baseos.qe:389

So it appears that "ad1.ad.baseos.qe" is still resolvable into IPv4 address? Can you remove the A record or change the address priority in sssd.conf to:
lookup_family_order = ipv6_first

and restart the sssd?
Comment 3 Patrik Kis 2013-10-21 11:53:33 EDT
Yes, this quite make sense. Adding lookup_family_order = ipv6_first to sssd.conf make it work. OK, so this is not really an sssd bug; sorry, I was confused by the fact that it started to pass after getting the credentials in ipv4 env. Although, sssd could fall back to other address automatically.

Or maybe realmd could add lookup_family_order = ipv6_first if it detect (if it can detect it) pure IPv6 environment.
Comment 4 Jakub Hrozek 2013-10-21 16:18:29 EDT
I think I will turn this into a bug against SSSD (although not the biggest priority): if both v4 and v6 can be resolved, try both v4 and v6.

I would first like to discover how other software performs this fallback, so this might end up being a docs bug, but I'd like to to not just close it.
Comment 5 Patrik Kis 2013-10-22 02:24:10 EDT
I agree. Feel free to decrease the priority; originally it looked like a more serious problem that's the reason for the high priority.
Comment 6 Jakub Hrozek 2013-10-24 06:21:40 EDT
Upstream ticket:
https://fedorahosted.org/sssd/ticket/2128
Comment 9 Jakub Hrozek 2016-01-08 07:18:14 EST
This bugzilla depends on larger failover/resolver changes which are not scoped for 7.3..
Comment 10 Jakub Hrozek 2016-03-02 05:15:16 EST
Upstream ticket:
https://fedorahosted.org/sssd/ticket/2015
Comment 11 Jakub Hrozek 2016-11-23 09:23:59 EST
Thank you for filing this bug report.

Since fixing this bug requires work in the upstream SSSD project which is not scheduled for the immediate future, I added a conditional development nack, pending upstream availability.

Note You need to log in before you can comment on or make changes to this bug.