#2357 Failover SRV discovery not honouring priority/weight
Closed: Fixed None Opened 5 years ago by kieren.

Using sssd-1.11.5.1-1.el6.x86_64 (from jhrozek's recent COPR repo) on RHEL6.5.

I have the following SRV RR's configured in DNS:

;; QUESTION SECTION:
;_ldap._tcp.idm.poc1.local.        IN      SRV

;; ANSWER SECTION:
_ldap._tcp.idm.poc1.local. 3600 IN SRV     0 0 389 ipa2.idm.poc1.syseng.tmcs.
_ldap._tcp.idm.poc1.local. 3600 IN SRV     0 0 389 ipa1.idm.poc1.syseng.tmcs.
_ldap._tcp.idm.poc1.local. 3600 IN SRV     10 0 389 ipa2.idm.poc2.syseng.tmcs.
_ldap._tcp.idm.poc1.local. 3600 IN SRV     10 0 389 ipa1.idm.poc2.syseng.tmcs.
_ldap._tcp.idm.poc1.local. 3600 IN SRV     20 0 389 ipa2.idm.poc3.syseng.tmcs.
_ldap._tcp.idm.poc1.local. 3600 IN SRV     20 0 389 ipa1.idm.poc3.syseng.tmcs.

Tcpdump confirms all 6 records are returned by the DNS server during SSSD's startup.

The SRV discovery only tries one host from each priority level, ie; 3 servers in the above case.
Debug logs seem to confirm this:

(Wed Jun 11 07:03:57 2014) [sssd[be[idm.local]]] [fo_discover_srv_done] (0x0400): Got answer. Processing...
(Wed Jun 11 07:03:57 2014) [sssd[be[idm.local]]] [fo_discover_srv_done] (0x0400): Got 3 servers
(Wed Jun 11 07:03:57 2014) [sssd[be[idm.local]]] [fo_add_server_to_list] (0x0400): Inserted primary server 'ipa2.idm.poc1.local:389' to service 'IPA'
(Wed Jun 11 07:03:57 2014) [sssd[be[idm.local]]] [fo_add_server_to_list] (0x0400): Inserted primary server 'ipa1.idm.poc2.local:389' to service 'IPA'
(Wed Jun 11 07:03:57 2014) [sssd[be[idm.local]]] [fo_add_server_to_list] (0x0400): Inserted primary server 'ipa2.idm.poc3.local:389' to service 'IPA'
(Wed Jun 11 07:03:57 2014) [sssd[be[idm.local]]] [set_srv_data_status] (0x0100): Marking SRV lookup of service 'IPA' as 'resolved'

Full debug log and sssd.conf attached.

RFC2782 suggests that for each priority level, all elements should be added to the list (ordered by weight).


Fields changed

milestone: NEEDS_TRIAGE => SSSD 1.12.1

Fields changed

owner: somebody => pbrezina
status: new => assigned

Fields changed

patch: 0 => 1

Fixed in master:
- 2be2220
- da6ee4b
- cb8c792

resolution: => fixed
status: assigned => closed

Reporter -- if you'd like to have the fix in RHEL-6 as well, kindly work with Red Hat support to open a bug against RHEL-6.

Thank you very much for reporting the bug.

Metadata Update from @kieren:
- Issue assigned to pbrezina
- Issue set to the milestone: SSSD 1.12.1

2 years ago

Login to comment on this ticket.

Metadata