【RAC】PRVF-4664/4657: Found inconsistent name resolution entries for SCAN name

涉及到的版本： 11.2.0.1 to 11.2.0.2 - Release: 11.2 to 11.2

现象：

After successful 11gR2 Grid Infrastructure clusterware (CRS) installation or upgrade, 

INS-20802 reported while running Oracle Cluster Verification Utility post check, installation logs located in $ORACLE_BASE/oraInventory/logs/installActions{$TIMESTAMP}.log shows similar entries like following:

INFO: PRVF-4664 : Found inconsistent name resolution entries for SCAN name "cluscan.us.oracle.com"

INFO: ERROR:

INFO: PRVF-4657 : Name resolution setup check for "scanclunm" (IP address: 10.4.0.202) failed

INFO: ERROR:

INFO: PRVF-4664 : Found inconsistent name resolution entries for SCAN name "cluscan.us.oracle.com"

If you execute "$GRID_HOME/bin/cluvfy comp scan", it shows similar error.

原因：

Cause 1. SCAN name is resolved by hosts file instead of DNS or GNS, this could happen even you list all SCAN IP(s) in hosts file

Cause 2. nslookup fails to find record for SCAN name:

nslookup cluscan.us.oracle.com

..

** server can't find eotcs.us.eot.com: NXDOMAIN

Cause 3. Other DNS hostname is canonical name(CNAME) of SCAN:

nslookup cluscan.us.oracle.com

..

cluscan.us.oracle.com     canonical name = cluscan.a3.oracle.com

Name:   cluscan.a3.oracle.com

Address: 10.4.0.201

Name:   cluscan.a3.oracle.com

Address: 10.4.0.202

Name:   cluscan.a3.oracle.com

Address: 10.4.0.203

Cause 4. DNS is configured properly but other naming resolution(nis, ldap..) is being used and doesn't have proper SCAN info

Cause 5. Persistent cache for nscd has incorrect information.

解决方法：

For cause 1: Oracle strongly recommend to use DNS or GNS for SCAN name resolution as hosts file support only one IP for SCAN name. 

If you intended to use hosts file for SCAN name resolution, and ping command returns correct SCAN VIP, you can ignore the error and move forward.

If you intend to use DNS or GNS for SCAN name resolution, please comment out entries in hosts file for SCAN name on all nodes, and re-run "$GRID_HOME/bin/cluvfy comp scan" to confirm.

For cause 2: Please engage System Administrator(SA) to check resolver configuration (/etc/resolv.conf on Linux/UNIX), correct any misconfiguration on all nodes and re-run "$GRID_HOME/bin/cluvfy comp scan" to confirm.

For cause 3: Please engage SA to update SCAN record in DNS to A type instead of CNAME type.

For cause 4: Please engage SA to check name switch configuration (/etc/nsswitch.conf on Linux, Solaris and hp-ux or /etc/netsvc.conf on AIX) and correct any misconfiguration on all nodes. Example hosts in nsswitch.conf:

hosts:      files dns nis

Once it's corrected, execute "$GRID_HOME/bin/cluvfy comp scan" to confirm.

For cause 5: Please engage SA to restart nscd and clear persistent cache on all nodes. Example on Linux

# /sbin/service nscd restart

# /usr/sbin/nscd --invalidate=hosts

Once it's corrected, execute "$GRID_HOME/bin/cluvfy comp scan" to confirm.

------------------------------------------------------------------------------------------------------------

What we expect to see when executing nslookup and ping

nslookup cluscan.us.oracle.com

..

Name:   cluscan.us.oracle.com

Address: 10.4.0.201

Name:   cluscan.us.oracle.com

Address: 10.4.0.202

Name:   cluscan.us.oracle.com

Address: 10.4.0.203

ping -c 1 cluscan.us.oracle.com

PING cluscan.us.oracle.com (10.4.0.201) 56(84) bytes of data.

64 bytes from cluscan.us.oracle.com (10.4.0.201): icmp_seq=1 ttl=64 time=0.258 ms

--- cluscan.us.oracle.com ping statistics ---

1 packets transmitted, 1 received, 0% packet loss, time 0ms

rtt min/avg/max/mdev = 0.258/0.258/0.258/0.000 ms

ping -c 1 cluscan.us.oracle.com

PING cluscan.us.oracle.com (10.4.0.202) 56(84) bytes of data.

64 bytes from cluscan.us.oracle.com (10.4.0.202): icmp_seq=1 ttl=64 time=0.258 ms

--- cluscan.us.oracle.com ping statistics ---

1 packets transmitted, 1 received, 0% packet loss, time 0ms

rtt min/avg/max/mdev = 0.258/0.258/0.258/0.000 ms

ping -c 1 cluscan.us.oracle.com

PING cluscan.us.oracle.com (10.4.0.203) 56(84) bytes of data.

64 bytes from cluscan.us.oracle.com (10.4.0.203): icmp_seq=1 ttl=64 time=0.258 ms

--- cluscan.us.oracle.com ping statistics ---

1 packets transmitted, 1 received, 0% packet loss, time 0ms

rtt min/avg/max/mdev = 0.258/0.258/0.258/0.000 ms

From above you can see:

  1. nslookup is returning proper SCAN name and IPs

  2. ping to SCAN name is Round Robin resolved to different IP each time by DNS

If you see different behaviour than above, please engage SA to investigate.

来自 “ ITPUB博客 ” ，链接：http://blog.itpub.net/22664653/viewspace-706838/，如需转载，请注明出处，否则将追究法律责任。

转载于:http://blog.itpub.net/22664653/viewspace-706838/