Conference smurf::ase

1973.0. "Unable to create ase in TCR" by NETRIX::"[email protected]" (John McDonald) Mon Mar 31 1997 16:56

I have a weird problem with an ase setup under TCR v1.4. The customer
had a SCSI cable problem, bu the FE fixed that, so now he's trying to
get the TCR/ase setup. Here's what we see:

The customer has installed TCR Production Server on two machines - a 4100
and a 4000. The install went OK (after we fixed the SCSI problem). He's
using virtual hub mode for the MC. Once both systems are booted, he
runs clu_ivp - everything shows up normal except for not having run asemgr
yet. He does a cnx_show command and sees what is expected, with the comment
about the tie-breaker disk not being defined.

The next thing he tries is to run asemgr to create the ase. He types in the
names of the two members MC interfaces (sid, slink), but asemgr tells him
there's an error and to check the logs. The following is what is showing up
in daemon.log:

Mar 26 16:10:10 Buzz ASE: local AseMgr Notice: Agent is in INSTALL STATE
Mar 26 16:10:38 Buzz ASE: local AseUtility Notice: ASE initial startup program
e
xiting...
Mar 26 16:10:39 Buzz ASE: local HSM Notice: Able to ping sid over the SCSI bus
Mar 26 16:10:44 Buzz ASE: local HSM Warning: Can't ping sid over the network
Mar 26 16:10:44 Buzz ASE: local HSM ***ALERT: HSM_PATH_STATUS:1.2.3.43:DOWN
Mar 26 16:10:53 Buzz ASE: slink AseMgr Error: new director hasn't started up
yet
...
Mar 26 16:10:57 Buzz ASE: local HSM Warning: network partition detected
between
local host and member sid
Mar 26 16:10:59 Buzz ASE: slink AseMgr Error: we're net partitioned from the
dir
ector
Mar 26 16:10:59 Buzz ASE: slink AseMgr ***ALERT: Net partition or disconnect -
c
an not find a director.
Mar 26 16:10:59 Buzz ASE: slink AseMgr Error: Unable to open the database.
Mar 26 16:12:38 Buzz ASE: local AseUtility Error: Can't fetch configuration
data
 base, fatal error!
Mar 26 16:12:38 Buzz ASE: local AseMgr Notice:
Mar 26 16:12:40 Buzz ASE: local HSM Warning: Can't ping sid over the SCSI bus
Mar 26 16:12:46 Buzz ASE: local HSM ***ALERT: HSM_PATH_STATUS:1.2.3.43:DOWN
Mar 26 16:12:46 Buzz ASE: local HSM Warning: member sid is DOWN
Mar 26 16:12:46 Buzz ASE: process_external_events: select error
Mar 26 16:12:49 Buzz ASE: local AseLogger Error: another AseLogger is running
on
 this node, exiting...
Mar 26 16:12:51 Buzz ASE: slink Agent Error: return from endless loop:Bad file
n
umber
Mar 26 16:12:51 Buzz ASE: slink Agent Notice: restarting Agent!
Mar 26 16:12:51 Buzz ASE: slink Agent Notice: in install state
Mar 26 16:12:56 Buzz ASE: local AseMgr Notice: Agent is in INSTALL STATE
Mar 26 16:12:58 Buzz ASE: local HSM Error: our_data_function: Error from write
d
ata ioctl
Mar 26 16:12:58 Buzz ASE: slink Agent Notice: in install state
Mar 26 16:13:07 Buzz ASE: local AseUtility Notice: ASE initial startup program
e
xiting...
Mar 26 16:13:07 Buzz ASE: local HSM Error: our_data_function: Error from write
d
ata ioctl
Mar 26 16:13:07 Buzz ASE: local HSM Warning: Can't ping sid over the SCSI bus
Mar 26 16:13:12 Buzz ASE: local HSM Warning: Can't ping sid over the network
Mar 26 16:13:12 Buzz ASE: local HSM ***ALERT: HSM_PATH_STATUS:1.2.3.43:DOWN
Mar 26 16:13:12 Buzz ASE: local HSM Warning: member sid is DOWN

At first I thought it might be a MC problem, so I had him bring down both
systems and run mc_diag and mc_cable - everything is fine. I then had him
ping each member over the MC and that worked fine. I told him to run
asesetup to re-create a new database and try it from the other machine -
the same problem. I logged onto his system and verified the hosts file -
each node had both MC interfaces and cluster_cnx defined the same and on
the same subnet.

Next I thought that it might be that he hadn't defined his tie-breaker disk
yet, so I had him do that. cnxshow then shows everything up and running,
and clu_ivp looks fine (except for not having an ase yet). I had him verify
the ASE_ID on both systems and they were the same (0).

At this point I thought he might have a corrupt install, so I had him
re-install TCR on both systems and rebuild the kernel. Tried to run
asemgr again and no luck - same error.

I checked his rc.config and svc.conf files and svcorder files, and everything
was as it should be.

Does anyone have any ideas as to what might be causing this, or where
I could look?

John McDonald
Atlanta CSC


[Posted by WWW Notes gateway]

    Hi John,
    
    >>  Buzz ASE: local HSM Warning: Can't ping sid over the SCSI bus
    
    
    Thus, it seams that the FE has not really fixed the SCSI BUS problem...
    
    Has the SCSI bus the same number on each system ?
    
    Are the SCSI ID different on each KZPSA ?
    
    Is the shared bus driven by supported SCSI controller (KZPSA) ?
    
    Regards, Manu.
    

John,

The first thing the HSM complains about is that the buzz can not ping  sid
over the network.  I would double check the MC configuration and make sure
it is ok.  There is nothing in the log which tells me that it is ok.

  Mar 26 16:10:44 Buzz ASE: local HSM Warning: Can't ping sid over the network

-- Greg
[Posted by WWW Notes gateway]