Apparently I'm breaking new ground here so I will share.

Solstice Disksuite is used to mirror and/or stripe disks, sort of a software
RAID.

Versions 4.2 and 4.2.1 support SNMP traps but need to be configured.

Once installed, there are three daemons that control SNMP.  Their control
scripts are stored in /etc/rc3.d.

They are S77dmi, S76snmpdx and S25mdlogd.  To determine their status, do a
ps -ef.  In all likelihood, the snmpXdmid is running.  Sun assures me that
this daemon is unnecessary for our needs (it works with their Sun Management
Centre).  Stop it by typing /etc/rc3.d/S77dmi stop.  Then rename S77dmi
s77dmi to prevent it from restarting on reboot.  

In all likelihood the mdlogd is NOT running.  It will not run unless it is
configured.  This is done using the mdlogd.cf file.  This is located either
in /etc/opt/SUNWmd (ver 4.2) or /etc/lvm (ver 4.2.1).  The file contains a
template as part of its comments.  Copy and paste this to the bottom of the
file and remove extraneous comments and leading spaces.

This should give you 5 lines.  Their example uses "spin" as the servername.
I will use "apollo", the name for my NT 4 server running What's up Gold 7.0.

ENTERPRISE =  1.3.6.1.4.1.42
OBJECTID =    1.3.6.1.4.1.860.1
"NOTICE: md:"   "apollo:162:udp"           6             1
"WARNING: md:"  "apollo:162:udp"           6             2
"PANIC: md:"    "apollo:162:udp"           6             3

The ENTERPRISE is Sun's generic OID.
The OBJECTID is Sun's 1.3.6.14.1 with 860.1 added to the end.  860 is unused
(and given in their example); I used 860.1 to make it unique (which may be
unnecessary).
The last three lines are notification messages.  Make sure you retain the
quotes.

Go to /etc/rc3.d and do the following.

./S76snmpdx stop 
./S25mdlogd stop
./S76snmpdx start
./S25mdlogd start

You may find yourself stopping and starting the snmpdx daemon a lot as it
seems to have problems on some machines.  The mdlogd daemon seems to work
independent of this so it may not be an issue.

In What's up Gold, under 'Configure', 'Notifications Library' create a
winpopup that displays somewhere where you can see it.  The 'message' should
be "%c %n %V %m %S(%s) at %t".  This is a modification of the default.  '%m'
gives you the text of the message.

You may also want to create a SMTPMail notification.  The 'mail server' is a
mail server that will accept incoming SMTP messages from your machine.  This
may take some negotiations with your mail server administrator.  The
'subject' should be "%c %n %m at %t".  Leave the rest as the default.

Next find (or create) the entry associated with your Solaris server.  Under
'properties', find 'services' and add the SNMP service (if you haven't done
so already).  Under 'Alerts', enable alerts then add an alert for Winpopup
and SMTPmail.  Set the 'notification' as appropriate and the 'trigger' to
9999.  Enable 'Send alert even if console response' and 'On SNMP Trap'.
Leave the box next to 'On SNMP Trap' empty for the moment.

Wait for one cycle to see if the SNMP is recognized.  If not, you may need
to stop and start the snmpdx daemon (see above).  Again, this may be
irrelevant.  You may need to add a Sun MIB (using mibextra) for this to be
recognized.

I ran my test on a Sparc5 (called 'karp') with an external drive chassis
containing 4 drives (SCSI ID's 0,2,4 and 5).  I created a mirror d0 with
concat/stripes c0t2d0s6 (d1) and c0t4d0s6 (d2) and a metadb spread over all
4 drives.  I took the cover off the drive bay so I could access the drives
while running.

I set up a console window and two terminal windows.  In one window, I cd'd
to /usr/opt/SUNWmd/sbin.  On the other window, I mounted /dev/md/rdsk/c0 to
/usr4.  I copied in a .tar file to make sure the mirror was working.

To actually perform the test, I pulled power to the drive on SCSI ID 2.  In
the second window, I untarred the file to give the drive something to do.
In the first window I did ./metastat until it noticed an error.  Fairly soon
I started to get console messages reporting "disk not responding to
selection", "WARNING: md: d1: /dev/dsk/c0t2d0s6 needs maintenance" and
"WARNING: md: d1: write error on /dev/dsk/c0t2d0s6".  The last two messages
were the ones the mdlogd daemon was looking for.

A moment later, on 'apollo' the What's Up Gold 7.0 server, I received two
winpopups containing the error message, time etc.  I also received two
emails at my workstation.  

----------------
Subject: Workstation karp Trap(sun-6.2) enterprises.860.1="Feb  7 09:02:59
2002

Workstation karp TRAP sun-6.2 at 09:04:24
Address: 10.10.10.10
Info 1: 
Info 2: 
Date: 02/07/2002
Status: Active and responding ( 0)
Svcs: 
Notes: 
Trap(sun-6.2) enterprises.860.1="Feb  7 09:02:57 2002
 karp unix: WARNING: md: d1: write error on /dev/dsk/c0t2d0s6"
----------------
and
----------------
Subject: Workstation karp Trap(sun-6.2) enterprises.860.1="Feb  7 09:02:57
2002

Workstation karp TRAP sun-6.2 at 09:04:25
Address: 10.10.10.10
Info 1: 
Info 2: 
Date: 02/07/2002
Status: Active and responding ( 0)
Svcs: 
Notes: 
Trap(sun-6.2) enterprises.860.1="Feb  7 09:02:59 2002
 karp unix: WARNING: md: d1: /dev/dsk/c0t2d0s6 needs maintenance"
----------------



Please visit http://www.ipswitch.com/support/mailing-lists.html 
to be removed from this list.

An Archive of this list is available at:
http://www.mail-archive.com/whatsup_forum%40list.ipswitch.com/

Reply via email to