This is my first posting to this group. I'vo got one problems and one
question. Thanks in advance for any help.
My confguration:
----------------------------------------
OS Platform: AIX 4.3.3
Tivoli NetView for UNIX for 6.0.2
Problem:
---------------
I've got following problem with false alarms in NetView.
Here is description of network and NMS configuration: There is a small
(about 16 devices) network based mainly on Cisco switches but there are 2
routers and 4 router cards in switch chasiss. Network is based on VLANs.
There is a separate management VLAN, where every device has its interface
in it. There is no DNS in this scenario. All names are defined in
/etc/hosts file. Because of safety reasons there is NO (!) routing on
routers between managment IP interface and other IP interfaces.Therefore I
cannot check interface status on routers using ICMP (ping), so i'm using
SNMP status poll feature to do this. It's configured through
/usr/OV/conf/netmon.seed file. IP addresess of these devices are registred
in this file, and prededed with "$" sign. Community names for all devices
are defined in "SNMP configuration" dialog. Devices are defined there by
names. There is additional configuration made in this dialog for every
device:
Status Polling: 3m
Timeout: 5.0
Retry count: 5
Configuration Polling interval: 1d
Problem: A few times a day I'm receiving false alarms about these devices.
There is a series of events: "Interface down", "Segment down", "Network
down"and finnaly "Router down", and router turns red on map. If I
immiedialty will perform demand poll action on this router, there is series
events "Interface up", "Segment up", Network up" and finnaly "Router up",
and router turns green on map. In will not perform "demand poll"action, the
same will happen after 3 mintes (what is status polling intreval defined by
me). Device examination shows that there were any real problems with this
device. Therefore I consider this alarm false.It simply seems that there is
no respond to SNMP get request from device. But in fact there are any
problems with communication with these devices. I can ping management
interface every moment (even immedialtely after false alarm) and response
time is less than 1 ms. Please note that there is 5s timeout, and 5 sec.
retry count for status polling, so there should be no problem with res
ponding to SNMP request.
What could be the reason of such a behavior. How can I track this?
Question:
---------------
My question relates to HSRP. Let's consider two scenarios:
1. There are two Cisco routers working as HSRP active and standby pair.
HSRP address is defined in netmon.seed (%). Due to current configuration
standby interface on standby router is disconnected. Active interface is
represented on map in active router as green. Standby interface is
represented in standby router as red. HSRP address is represented as
generic device in network segment as green.
2. There are another two Cisco routers working as HSRP pair. But their HSRP
address in NOT defined in netmon.seed. Both active and standby interfaces
are OK. Standby interface is represented in active router as green, standby
interface is represented in standby router as administrativly down, and
HSRP address is represented in active router as green.
I'm quite confused because I don't know which NetView behavour, HSRP
representation is correct. I thought that second one, ... but in this case
HSRP is not defined. AFAIK NetView could recognize HSRP interface by
specific MAC. Maybe it took place in this case (how such MAC should look
like). If second HSRP representation is correct - then what's wrong with
first one?
Pozdrowienia/Regards
Bartlomiej Grenda
_____________________________________
IBM Poland, I/T Specialist
tel +48-61-8649493, fax +48-61-8649488
e-mail: Bartlomiej.Grenda@pl.ibm.com
|