I've been trying to get LDAP authentication and NFS exported home directories on CentOS 6 working for a few days now. I've gotten to the point that I can now login to the client machine using the username and password in LDAP. On the client, /home and /opt are mounted in the fstab over NFS. However, every file in both /opt and /home is owned by nobody:nobody (uid: 99, gid: 99) on the client.
However my uid and gid appear to be properly set:
-bash-4.1$ id
uid=3000(myusername) gid=3000(employees) groups=3000(employees)
What else can I check? Here are some the config files on my client:
/etc/nsswitch.conf
passwd: files sss
shadow: files sss
group: files sss
hosts: files dns
bootparams: nisplus [NOTFOUND=return] files
ethers: files
netmasks: files
networks: files
protocols: files
rpc: files
services: files
netgroup: files sss
publickey: nisplus
automount: files ldap
aliases: files nisplus
/etc/sssd/sssd.conf
[sssd]
config_file_version = 2
services = nss, pam
domains = default
[nss]
[pam]
[domain/default]
auth_provider = ldap
ldap_id_use_start_tls = True
chpass_provider = ldap
cache_credentials = True
krb5_realm = EXAMPLE.COM
ldap_search_base = dc=mycompany,dc=com
id_provider = ldap
ldap_uri = ldaps://server.subdomain.mycompany.com
krb5_kdcip = kerberos.example.com
ldap_tls_cacertdir = /etc/openldap/cacerts
# Configure client certificate auth.
ldap_tls_cert = /etc/openldap/cacerts/client.pem
ldap_tls_key = /etc/openldap/cacerts/client.pem
ldap_tls_reqcert = demand
/etc/fstab
/dev/mapper/vg_main-lv_root / ext4 defaults 1 1
UUID=4e43a15d-4dc0-4836-8fa6-c3445fde756c /boot ext4 defaults 1 2
/dev/mapper/vg_main-lv_swap swap swap defaults 0 0
tmpfs /dev/shm tmpfs defaults 0 0
devpts /dev/pts devpts gid=5,mode=620 0 0
sysfs /sys sysfs defaults 0 0
proc /proc proc defaults 0 0
storage1:/nas/home /home nfs soft,intr,rsize=8192,wsize=8192
storage1:/nas/opt /opt nfs soft,intr,rsize=8192,wsize=8192
authconfig output:
[root@test1 ~]# authconfig --test
caching is disabled
nss_files is always enabled
nss_compat is disabled
nss_db is disabled
nss_hesiod is disabled
hesiod LHS = ""
hesiod RHS = ""
nss_ldap is enabled
LDAP+TLS is enabled
LDAP server = "ldaps://server.subdomain.mycompany.com"
LDAP base DN = "dc=mycompany,dc=com"
nss_nis is disabled
NIS server = ""
NIS domain = ""
nss_nisplus is disabled
nss_winbind is disabled
SMB workgroup = ""
SMB servers = ""
SMB security = "user"
SMB realm = ""
Winbind template shell = "/bin/false"
SMB idmap uid = "16777216-33554431"
SMB idmap gid = "16777216-33554431"
nss_sss is disabled by default
nss_wins is disabled
nss_mdns4_minimal is disabled
DNS preference over NSS or WINS is disabled
pam_unix is always enabled
shadow passwords are enabled
password hashing algorithm is sha512
pam_krb5 is disabled
krb5 realm = "EXAMPLE.COM"
krb5 realm via dns is disabled
krb5 kdc = "kerberos.example.com"
krb5 kdc via dns is disabled
krb5 admin server = "kerberos.example.com"
pam_ldap is enabled
LDAP+TLS is enabled
LDAP server = "ldaps://server.subdomain.mycompany.com"
LDAP base DN = "dc=mycompany,dc=com"
LDAP schema = "rfc2307"
pam_pkcs11 is disabled
use only smartcard for login is disabled
smartcard module = ""
smartcard removal action = ""
pam_fprintd is enabled
pam_winbind is disabled
SMB workgroup = ""
SMB servers = ""
SMB security = "user"
SMB realm = ""
pam_sss is disabled by default
credential caching in SSSD is enabled
SSSD use instead of legacy services if possible is enabled
pam_cracklib is enabled (try_first_pass retry=3 type=)
pam_passwdqc is disabled ()
pam_access is disabled ()
pam_mkhomedir or pam_oddjob_mkhomedir is enabled ()
Always authorize local users is enabled ()
Authenticate system accounts against network services is disabled
An note to add to this for google searchers - we had the same issue where no matter what we did, the nfs mount would not map the user ids correctly.
The issue was idmapd had cached the incorrect ids from the faulty configuration, and no fixing of the configuration would sort it.
The command on centos to fix this was nfsidmap -c (clear cache).
Hopefully this helps some desparate searcher..
Solved!
I happened to notice this line in
/var/log/messageson my NFS server when I was attempting to mount an export from the remote client:This caused me to look at the first few lines of
/etc/idmapd.conf:I then added
Domain=subdomain.mycompany.comunder the commented out "Domain" line. Saved, exited and then ran/etc/init.d/rpcidmapd restartand/etc/init.d/nfs restart.Is your NFS server running Centos/RHEL 5 by any chance?
If so, it is exporting NFSv3. NFSv4 is now the default for Centos6 (and recent Ubuntu variants).
The quick fix is to add "vers=3" in the mounting options in /etc/fstab .
e.g.
//10.0.0.1:/home /home nfs defaults,vers=3,rw,noatime 0 0
I found a blog post that might resolve your issue: http://whacked.net/2006/07/26/nfsv4nfs-mapid-nobody-domain/ which I found from the following forum post: https://www.centos.org/modules/newbb/viewtopic.php?topic_id=32977
Everything being mapped to "nobody" sounds like all_squash is turned on.
Take a look at:
http://www.centos.org/docs/5/html/Deployment_Guide-en-US/s1-nfs-server-config-exports.html
and verify that the NFS server's /etc/exports file doesn't unintentionally squash the UIDs. "no_all_squash" is supposed to be default, but you can try explicitly setting it and seeing what happens.
The fix for me is to make sure the DNS record exists for the local machine. Also helps if the reverse lookup record also exists. As a result, the nobody user and group was replaced by root. How simple is that?!? P.S. remember to reboot the local machine once the DNS records are created.