Veritas-bu

[Veritas-bu] SSO backup leading to status 83 and 213's on others

2002-04-15 10:10:51
Subject: [Veritas-bu] SSO backup leading to status 83 and 213's on others
From: gdrew AT deathstar DOT org (George Drew)
Date: Mon, 15 Apr 2002 10:10:51 -0400 (EDT)
DEVICE_HOST entries are only used for the java gui, and do not affect the
operation of media management in any way. If you have those entries on a
media server that means you are running the gui on that server, rather than
from the master. This is not unsupported, but I wouldn't recommend it, as
it's easier to mess something up that way.

Pg. 305 of the Media Management System Administrator's Guide explains the
MH_HOST_NAME setting. It is the name the local host will use to
register/reserve/release drives with the DAHost. Unless you're doing
something odd you should probably leave this alone.

The Global Database hostname should (typically) be your DAHost. Leave off
the -h option for vmglob. It should be set the same for all SSO hosts.

George

On Sun, 14 Apr 2002, Markham, Richard wrote:

> When I run a SSO backup it will complete successfully but it seems at some
> point afterwards i start getting media open errors (83) and then status 213.
> The problems reports will show permission denied entries as well.  I have
> been able to clear things up by commenting out the DEVICE_HOST entries on
> the media managers and also commenting out the DEVICE_HOST entries on the
> Master server for the media managers as well (leaving its own DEVICE_HOST
> entry untouched).  After I restart services on all media managers and master
> things start to run over the lan with no problems.
>
> I have one robot that has two multihosted drives and I add these drives to
> each of the media managers, making the master server the robot controller
> and volume database host.  i do a quick tldtest to make sure the mounts
> correspond to the expected drive #.
>
> On the media managers for this type setup should the vm.conf have a
> DEVICE_HOST = <hostname> or should it point to the Master?  Also for the
> media managers should there be a MH_HOST_NAME which points to the Master or
> should it be left to default to <hostname>?
>
> in regards to the two media managers i have configured (not including
> master):
> I did notice that I ran a SSO backup from one and it never caused a problem
> afterwards.  the configuration files match but when i stopped services on
> the second media server it stated vmd process was still active (this was
> several hours after a backup had completed and others had ran).  normally
> the media manager, while idle, will exit completey without prompting me to
> confirm the kill of the vmd process.
>
> i have trouble with 'Could not get globalDB hostname' with this same host.
> After doing a vmglob -set_gdbhost <master> -h <master> it will still return
> the same error when i vmglob -get_gdbhost.  i'm not sure if this is critical
> for the vmd to vmd/DA process or not but i believe its an issue and maybe
> vmd cant communicate as needed to successfully release share drive
> resources.
>
> thanks for any information.
>
>
>
>
>


<Prev in Thread] Current Thread [Next in Thread>