Our current setup is a Networker server running 7.3.2 on a Solaris 8 platform with storage node running 7.3.2 on a Solaris 9 platform. Almost all clients are running the same version. We have a few cluster clients where the physical node has been upgraded, but the virtual node still reports the old version.
Several times our server has just hung. At first there were GSS authentication errors reported. We disabled GSS and the same thing happened this weekend. Three backups were running and just stopped. There were no errors reported, below is the section of the daemon.log:
11/26/06 02:38:07 nsrd: stacker3:index:drnotes1 done saving to pool 'IndxSt2Weekly' (A05038) 502 MB
11/26/06 02:38:27 nsrd: stacker3:index:bhnsm02 saving to pool 'Weekly' (A06435L2)
11/26/06 02:43:03 savegrp: job (53779) host: f68vrec savepoint: /u01 had WARNING indication(s) at completion.
pools supported: IndxSt2Weekly;
11/26/06 03:50:09 savegrp: Aborting inactive job (53513) lnotes3:/plaza2
11/26/06 03:50:10 savegrp: command 'save -s stacker3 -g fullweekly -LL -f - -m lnotes3 -l full -q -W 78 -N /plaza2 /plaza2 ' for client lnotes3 exited with return code 1.
11/26/06 03:50:10 savegrp: job (53513) host: lnotes3 savepoint: /plaza2 had WARNING indication(s) at completion.
11/26/06 03:50:10 savegrp: lnotes3:/plaza2 failed.
* lnotes3:/plaza2 (interrupted), exiting
* lnotes3:/plaza2 aborted due to inactivity
11/26/06 03:50:10 savegrp: lnotes3:/plaza2 will retry 1 more time(s)
11/26/06 04:03:05 nsrlcpd #1: Jukebox `lto1' is exiting. The jukebox is no longer managed by nsrlcpd.
Then there was nothing in the daemon log and nsrwatch got hung as well. I could not even do an nsr_shutdown. I had to kill all processes one at a time. Then when I restarted the daemons, there was an error saying the volumes have reached the low water mark. A second restart of the daemons cleared that.
GSS authentication is a new method of authenticating, it uses ssl.
Here is how we disabled it:
# nsradmin -p nsrexec NetWorker administration program. Use the "help" command for help, "visual" for full-screen mode. nsradmin> . type: NSRLA (note: there is a blank space between . and type) Current query set nsradmin> show auth methods nsradmin> print auth methods: "0.0.0.0/0,nsrauth/oldauth"; nsradmin> update auth methods: "0.0.0.0/0,oldauth" auth methods: "0.0.0.0/0,oldauth"; Update? yes updated resource id 3.0.156.101.198.4.188.67.137.69.101.64(137) nsradmin> print auth methods: "0.0.0.0/0,oldauth"; nsradmin> quit
This has happened several times before. The first couple times there were GSS authentication errors reported. After disabling GSS, there were no errors reported.
Funny enough, today I was playing with that on HPUX server and did manage to break index backups (that would be fine after restart). After few tweaks I managed to make it work, but it's really amazing how unstable can be.
Jodi, the GSS is disabled on the NW server side or on the NW client/storage node side ? Suppose to have a client in v7.2.1, a storage node and a server in v7.3.2. I have to
1) do nothing on client side ( v7.2.1 has no authentication method ) 2) modify storage node configuration with noauth 3) modify server configuration with noauth
It looks like we're in the clear. After upgrading the hardware all problems have gone away. We have not re-enabled GSS and probably won't. It doesn't seem necessary.
ble1
6 Operator
•
14354 Posts
•
56186 Points
477
0
Posted November 27th, 2006 07:00