[Freeipa-users] Kerberos hanging

[Terry John]

I have a problem using freeipa version 3.0.0-50 on CentOS release 6.8. The problem manifests itself as no authentication, and no DNS.

It seems Kerberos just stops responding to requests and requests just get queued up
# netstat -tuna | grep SYN_RECV
Active Internet connections (servers and established)
Proto Recv-Q Send-Q Local Address               Foreign Address             State
tcp        0             0           <server IP>:88               <client1 IP>:55440         SYN_RECV
tcp        0             0           <server IP>:88               <client 2 IP>:40076        SYN_RECV
tcp        0             0           <server IP>:88               <Client 3 IP>:41525        SYN_RECV
tcp        0             0           <server IP>:88               <Client4 IP>:53958         SYN_RECV
tcp        0             0           <server IP>:88               <Client5 IP>:54240         SYN_RECV

Looking at /var/log/krb5kdc.log
The normal activity of AS_REQ and TGS_REC messages just stops. No error messages. Just  no new messages.

In /var/log/messages the named server reports messages like
Mar  1 00:00:23 freeipa named[18989]: LDAP error: Can't contact LDAP server
Mar  1 00:00:23 freeipa named[18989]: connection to the LDAP server was lost
Mar  1 00:00:23 freeipa named[18989]: bind to LDAP server failed: Can't contact LDAP server

The command "kinit" is totally unresponsive and will time out if you wait long enough.

If I try to restart ipa with "service ipa restart", it hangs on the first stage, trying to stop DIRSRV.
I have to do a "ps ax" and look for this line.
2758 ?        Sl     2:13 /usr/sbin/ns-slapd -D /etc/dirsrv/slapd-MY-REALM -i /var/run/dirsrv/slapd-MY-REALM.pid -w /var/run/dirsrv/slapd-MY-REALM.startpid

Then I have to a "kill -9" on the pid
Then I can do "service ipa restart"

After that it works ok for a while. "A while" may be a few minutes or several hours.
The filesystem is only 58% used and "free" shows no swap in use so there seems to be plenty of RAM available.
"top" shows CPU(s) 96% idle with "dirsirv" typically using about 3%CPU at most

I've no idea why this keeps happening, everything looks ok then it just stops

Terry John
System Administrator- Cox Automotive Software
E: Terry.John at coxauto.co.uk



Cox Automotive group of companies within the UK comprises: Cox Automotive UK Limited (registered number: 03183918), Manheim Limited (registered number: 00448761), Cox Automotive Retail Solutions Limited (registered number: 02838588), Motors.co.uk Limited (registered number: 05975777), and Real Time Communications Limited (registered number: 04277845). Each of these companies is registered in England and Wales with the registered office address of Central House, Leeds Road, Rothwell, Leeds LS26 0JE. The Cox Automotive group of companies within the UK operates under various brand/trading names including Cox Automotive UK, Manheim Inspection Services, Manheim Auctions, Modix, Xtime and Closeit.

V:0CF72C13B2AD