Summary: A rolling patch on a one member cluster.

From: Jim Fitzmaurice <jpfitz_at_fnal.gov>
Date: Thu, 08 Feb 2001 13:12:16 -0600

Hello,

    Well the list did have a few tips, and thanks to all who commented. Most
of the responses I got were from people having similar problems. I ended up
going to Compaq for the answer, it was amazingly simple, although it took 2
days to figure it out. The worst part is I had all the info, I just didn't
put two and two together, so now I get to kick myself for not figuring this
out on my own.

    The problem was I had cfgmgr commented out in my /etc/inetd.conf file. I
removed the comment (#) and "HUP'ed" my inetd and everything worked
perfectly.

My original problem is below.

Jim Fitzmaurice
jpfitz_at_fnal.gov

UNIX is very user friendly, It's just very particular about who it makes
friends with.

----- Original Message -----
From: Jim Fitzmaurice <jpfitz_at_fnal.gov>
To: tru64 <tru64-unix-managers_at_ornl.gov>
Sent: Tuesday, January 30, 2001 1:45 PM
Subject: A rolling patch on a one member cluster.


> TruCluster managers,
>
> I'm new to TruClusters and I'm building a TruCluster 5.1 system
starting
> with an GS80 running Tru64 V5.1. I only have the one member at this time
as
> I haven't added the others yet. I wanted to put on patch kit 2 before
adding
> the other members. I tried to do a straight install since this was the
only
> member currently in the cluster but I got the message:
> --------------------------------------------------------------------------

--
> --------------------------------
> Checking Cluster State...done
>
> This system is part of a V5.0 cluster which has not been prepared to do a
> rolling patch installation. Refer to the Patch Installation Guide as to
the
> proper procedure to start a rolling patch.
> --------------------------------------------------------------------------
--
> --------------------------------
>
> I figured O.K. no problem, I could do that. So I started doing a rolling
> patch, figuring it would be good practice, but the first command,
> "clu_upgrade -v check setup 1" gave me the following error:
> --------------------------------------------------------------------------
--
> --------------------------------
> Retrieving cluster upgrade status.
> Retrieving cluster status.
> Retrieving upgrade-related system configuration.
>
>
> *** Error ***
> All members must be running without tagged files.
> --------------------------------------------------------------------------
--
> --------------------------------
>
> Well, I couldn't get any of the "clu_upgrade tagged" commands to work at
> all, They all gave me the following error:
> --------------------------------------------------------------------------
--
> --------------------------------
> *** Error ***
> Member '1' appears to be booting or is in single-user mode. Either wait
for
> the node to continue booting, boot the node to multi-user mode, or halt
the
> node using the shutdown command. After all members are running in
multi-user
> mode or halted, you can rerun the clu_upgrade command.
> --------------------------------------------------------------------------
--
> --------------------------------
>
> Now this was confusing because this system, Member '1' was up and in
> multi-user mode. A "who -r" command showed the system to be in run-level
3.
> Undaunted I decided to search the system with the "find" command to locate
> any ".Old.." files. The command found none anywhere on the system. I
figured
> the error I got on the first clu_upgrade was just a warning message not a
> show stopping error. So I continued with, "clu_upgrade setup 1" and got
the
> same error the "tagged" commands were giving me.
> --------------------------------------------------------------------------
--
> --------------------------------
> *** Error ***
> Member '1' appears to be booting or is in single-user mode. Either wait
for
> the node to continue booting, boot the node to multi-user mode, or halt
the
> node using the shutdown command. After all members are running in
multi-user
> mode or halted, you can rerun the clu_upgrade command.
> --------------------------------------------------------------------------
--
> --------------------------------
>
> O.K. this appears to be a show stopping error. I ran clu_get_info to make
> sure and yes this is definitely Member '1', currently it's the only
member.
> What could be wrong, why doesn't it see itself in run-level 3? Can anybody
> help? I've been through the books and they are very good it explaining to
> you how it's suppose to work, but don't give you much help when something
> goes wrong.
>
> Jim Fitzmaurice
> jpfitz_at_fnal.gov
>
> UNIX is very user friendly, It's just very particular about who it makes
> friends with.
>
>
>

Received on Thu Feb 08 2001 - 19:01:18 NZDT