OpenNMS 101 - Module 10: Passive Status Keeper Video Lecture | A Complete Guide: OpenNMS - IT & Software

hi gang and welcome to another episode
of open an MS 101 we're getting toward
the end of the class here and this has
got to be my favorite lesson while it
focuses on the passive status keeper it
also includes things like the event
translator and in this lesson we really
tie together a lot of the concepts that
we've been discussing throughout this
class so if you haven't seen the other
modules you may get a little lost in
this one assume you have and if you have
then you're ready to go so the passive
status keeper what is it well we were
approached many years ago by a company
that did telephone television
transmission in Australia so big
television company in Australia and of
course they used a lot of satellites now
the problem was satellites don't
necessarily have IP addresses they're
not you're not able to get information
out of them so our normal method of
service assurance be active polar where
we actually go out and test something on
the network failed in this scenario but
they did add this box it was
manufactured by a company called pixel
metrics and pixel metrics would send
events to a management station like open
NMS it could send the Venson to open in
a mess that would relate to the status
of these satellite dishes so let's say a
transceiver was down or some part of the
system was down it could send a trap
when that part of the system was fixed
it was in another trap and the idea was
could we map the open ms model of having
like a node interface in the service to
something that wasn't this active
service that we actually went out and
monitored directly and so we created
this thing called the passive status
keeper it's a way of tracking status for
something that we can actively pull and
then that status is changed with an
event so basically the way it works is
quite simple there's a table in memory
and on that table there is a node and
interface in service just like we have
now
and whether it's up or down so there's
only there's four things in the table
one per each entry now when open NMS
starts up it goes through its
configuration in the database and says
oh these nodes these interfaces have
these passive services so it adds this
to this table in memory and it sets the
status of course to up and then it looks
in the outage table and says do any of
these passive services have an active
outage if so the outage is the status of
those services is marked two down and
that's pretty much it now we have this
event that basically will come in and we
can set this event to change the status
of any one of these particular services
but the actual service check the monitor
is is quite simple it just goes to
memory and says hey is there a you know
what's the status of this service if
it's down it generates all the new bla
service events and things that we're
used to if it's up nothing happens makes
sense now how do we change this by
events well there's another really cool
feature in open NMS called the event
translator and what the event translator
does is it takes one event and it
converts it into another event there's a
lot of uses for this and some of this is
event enrichment we have a number of
customers who will add their own
database tables to the open unless
database and basically they might have
you know which customers are affected by
a particular service running on a node
and so if that service goes down if it's
an interface on a router if it's a web
server etc when the node loss service
event comes in the event translator can
take that translate that into an
enhanced event in which in say the
operator instruction field or one of the
other fields of the event you can put
hey this is the impact here are the
customers that are impacted by this so
it's really really kind of cool
now you can change the whole event or
you can inherit a bunch of stuff from
the original event you have to change
the unique event identifier the UI
otherwise you don't want to have a loop
as it will send it back in and come back
in I'm not even sure if it will actually
run if you have the uei of the
triggering event to be the same as the
translated event so for the passive
status keeper the PSK we actually take
an event and create we turn it into a
passive service status event and we'll
let me show you how to do that now I'm
going to jump back over here to my my
terminal window and I'm going to edit
the translator configuration XML file so
the translator configuration XML file we
ship a number of translations with open
NMS one of the most useful ones is SNP
link down limit now we talked about SMT
in earlier lessons and there's a very
very standard very common event called
Lake mail basically if you go to a
device you pull the cord out if the link
light goes from on to off that usually
will generate a link down message now in
the generic trap in the traffic that is
specified by the SMT standard there's
only one variable set with that trap and
that is the index and so how useful is
that either you get a message goes oh no
index seven is down no not in the next
seven so the idea is can we enrich these
and make them more useful in the fact
matter is yes basically in the database
we have a number of information for a
particular IP address and an interface
and as an if' index and those things
include the interface description the
interface name and the interface alias
all which we have gathered and put in
the database via SNMP so let's look at
this translation here so the triggering
event is this uei right here let me get
rid of that highlight so we have this
event right here UVI dot open msl to
work slash generic slash traps SMP link
down so when that event happens it's
going to trigger this event translation
and so the first thing we do is we set
at the UVI of our new event to pretty
much the same thing except instead of
saying generic traps like it says here
we put translator traps and if you
remember the unique event identifier is
a string that uniquely identifies a
particular event so by changing this
from generic to translator we are able
to actually will generate a new a
totally new event now notice this
assignment name this is just a field so
basically we're going to take a field in
the original event and we are going to
change it by setting it to a constant
which is just this string so we we hard
code this straight into the EEI field
now the next thing we're going to do is
we're going to add three parameters now
as you remember events can have
parameters and so we're going to add
three new parameters we're going to call
one of them on the interface description
one of them the interface name and one
of them the interface alias now to get
these informations we actually have to
run an SQL statement and so here's the
SQL statement we're going to do we're
going to say select the SP interface
description from the SPI interface table
where the note ID equals question mark
and the SVT if' index equals question
mark now these little colon colon
integers they're just required for the
link we have to the database but the big
thing is to notice these question marks
what do you say is we need to fill these
in with certain variables and so what
we're going to do is going to pass into
this SQL statement a very the first
variable which is going to be the node
ID field that came with the original
event
so the aspecting link down event comes
in we're going to set this first
question mark here to be whatever comes
back from this is matches this is a
regular expression and we'll set the
result to be with the whole thing
whatever comes back from the node ID
field so now our let's say this was note
17 so this would say select SP from SP
SP ifs occur from SP interface where
our SMT no 90 equals 1717
now the next thing we have to get is the
index now it turns out the index comes
from a parameter set from the first
event so we have a link down event comes
from the sentence parameter and of
course we have to do all these weird
escapes in default SNMP the parameters
are named by the object ID so in when we
create this parameter here we're going
to call it is desk er but what's going
to happen is that for the original event
because it was an SNMP trap the variable
that gets in the VAR bind is actually
going to have an object ID and it's
going to start out down 1.3 that 6.1.2
dot one dot two dots do dot one dot one
dot something
so dot one two three to six dot one
that's internet so that's org I so org
department of sends Internet to is
management one is nib to two is the
interface table two is the two is
interface branch to is interface table I
accentuated first entry is I F index and
then the if' index is going to be
whatever comes after here now I hope
that you guys have some experience with
regular expressions if not feel free to
use some of our examples of the template
but basically these little question
marks there
excuse me parentheses or groupings so
basically what this says is starting
with so this regular expression is the
tilde character carrots is starting with
find the parameter that starts with 1.3
that 6.2 don't 1.2.0 1.1 dot and then
collect anything after that
and then this dollar sign says the in so
you're looking for the end of the
parameter name so anywhere from this dot
after this dot to the end when collect
that number which happens to be the I've
index and so we'll plug that so this is
eyes at the next one so now our
statement would read select the SP
interface description from the SDP
interface table where note ID equals 17
and the SPI index equals 1 and so that
will return if it exists it will
turn in the description and then we
repeat that for the name and we repeat
that for the alias so basically what we
end up doing here is we end up adding
three fields three parameters to our
event and then we send the translator
event into open and a mess now let me
look at this let me grab this UVI for a
second if you remember all the events
are defined an event file and I'm if I
edit the default events I'm a search for
this guy so let me do it do you can do
they don't escape so there is the
generic link down right here and you can
see we set to degree vent labeled log
message notice we set this to do not
persist so what do not persist does is
this will not store this original link
down in it anywhere so when we store in
the database instead we'll deal with the
translated event so we'll never see that
the original link done event will only
see the translated vent so if you're
using the event translator you pretty
much want to either hide either via do
not persist or via setting this to log
only you want to kind of hide the event
so that it's not shown to the end-user
so the image user can focus on the
translator event which will have all the
interesting and useful information that
makes sense so if we look let me go back
up a directory go back into the
translator huh go back into the
translator so you can see we have one
for SVP link down we have a topology
link down event this is very very
similar we have the link up event we
have one for high Pyrrhic we have a
cisco configuration management again all
of these things we just kind of look up
and add various and different things to
these events to make them more useful
now what we're going to end up doing is
for the past the status keeper we're
going to end up doing adding an event
translation and I just want to stress
that the event translator could you know
it stands on its own to
you can use it you don't have to use the
passive status keeper and they've a to
use these that translator the event
translator as you saw with the link down
event you can enrich events and this is
huge there's there's a number of
expensive commercial products that
measure impact and basically depending
on how much information you have about
your network and how instrumented it is
you can actually use this event
translator to provide very detailed
information to your network operators
okay so how do we deal with the how do
we deal with the PSK we're going to
create a new for this exercise when
create a new service called class mood
now if you remember when we talked about
SNMP traps we created a class's Moody
and a class's joyful trap so what we're
going to do here is we're going to use
those traps to trigger this passive
class made service so we're going to do
is we're going to create a class mood
and in the last lesson I talked about
adding services to a particular node in
this case we're going to do it ad-hoc
we're not going to create a detector the
detects classroom then be actually very
very difficult if you think about it
because you have to have some way to
know that a passive service needs to
have that the passive service needs to
exist on this particular interface and
to me that's you know if it's passive
that became a difficult thing of a few
ways to do it but I think it's kind of
difficult so we're going to do is we're
going to add this class mood service to
our local host node will synchronize the
requisition and then we'll notice that
we'll look at the monitoring status I'll
give you a big spoiler it's going to say
not monitored because we haven't added
it to any kind of polling package so
they want to add this passive status
keeper monitor to our default polling
package we're going to update the event
translator to take our traps our Moody
and joyful traps and turn them into
passive status events
then because we monitor to modify the
polling package we're gonna have to
restart open NMS so we'll create an
outage and we'll see what happens and
then
or the outage make sense if not hit
rewind or hold on for a few seconds and
we'll get through this exercise okay so
I'm going to jump back over here to my
GUI so here is my GUI Lani we enlarge
that so it takes up the whole screen so
here we are there's no outages there's
no alarms everything it's nice and green
so that's kind of hold so I'm going to
go over to configure open and a mess
managing provisioning requisitions and
I'm going to edit our class requisition
by clicking on the little pencil and
then I want to edit my local host node
and then we have these interfaces we
have this one $27.00 another one I'm
going to edit that guy and I'm going to
add a service now notice it says service
name choose one from this drop-down list
or type in custom text for new ones so
I'm just going to type in class mood now
I know I've stressed this a number of
times but repetition is basic to all
learning so I'll repeat it
climb the service name you choose here
is important because it's going to
percolate throughout everything and so
don't use spaces in this case I use the
- I could use an underscore character or
a friend of mine used to say underbar
I'm going to use an underscore character
or camel case those are the best ways to
do it
so I'm going to just I've done this I've
just manually added to serve someone hit
save once this service is added to a
node it will be available on a drop-down
box so if you say want to put this class
mood service on a number of devices
can't imagine why but if you did you
could go and choose it from the drop
down once it is associated with a
particular mode so I'm going to save
this hit return and now I'm back here I
will return again and notice that my
requisition is red because I've made
changes and I'm going to hit synchronize
I'll hit yes so basically we just
synchronized everything goes white which
says all changes have been applied and
I'm going to go to my note page and as
you can see on the one 27.1 interface we
now have this class mood service and as
promised
it says not
long injured make sense okay let's get
back to the slides so we've added the
classman service it's not monitored so
we're going to create this this monitor
service in the polar configuration it's
one line
remember how especially with the last
lesson when we did the page sequence
monitor it was a huge amount of XML here
this just says service name class mood
the interval to check is 30,000
milliseconds or 30 seconds user fine
equals to remember that's not really
important but I try to use it anyway and
then status equals on that's it that's
why we have this little slash greater
than sign because there's nothing else
there if there was any service that you
wanted to monitor at a millisecond rate
this would probably probably be it
because all it's doing is checking
memory so it's very very fast I just
said it for 30 there's no reason I said
it much faster than that and plus it'll
allow me to demonstrate some of the
up-and-down things won't create now
remember when you actually do this you
actually have to define what software
basically what class the service runs
now I made a mistake here this is
service mood it should say class special
mood and it does say class that's
removed in the in the screen that I have
to update these slides before I publish
them so yeah this should say class -
movie because again it counts if I just
left this as it is here it when match so
I'll fix it in this slide when I upload
these today to the website but notice it
uses the passive service monitor that is
the name of the monitor latency that's
the status keeper okay now here is our
now we've added the service again all
these service check does is look at that
table in memory and goes it up in the
deck is it up is down boo boo boo boo
boo now in order to make it go up and
make it go down we have to generate a
passive service status event and to do
that we're going to use the event
translator now for simplicity for
simplicity I've made this fairly simple
always basically done is use constants
now in reality if you were using this on
multiple systems you would have to have
some way of like looking up what the
particular what the particular hostname
and IP address etc in the case of this
television provider in Australia what we
did is we created virtual nodes we
basically provision the node that was
like satellite one satellite two
satellite threes I'd like four etcetera
and then we add a satellite one was
10.1.1.1 satellite two was 10.1.1.10 we
just made up IP addresses and then we
could look those up in the database when
these events came in but rather than
complicate this I kind of wanted just to
put this in here as as simply as
possible so we're going to take the UV
ID open em and stored class moody event
that's going to trigger this and then
what we're going to end up doing is
we're going to set there's the passive
the passive status service status event
requires three to four parameters or
nobody expects the specs for parameters
so we've got the passive node label the
passive IP address the passive service
name and what is their status is it up
or down remember these four things
correspond directly to the table that's
a memory there's a node label or host ID
passive service IP address the name and
the status so we're going to do is say
when the Moody is that when the Moody
trap comes in it'll get converted into
an event the Moody event will trigger
this translation and we'll set the node
label parameter to the hard-coded
constant localhost hard-coded one 27001
hard-coded service class food and
hard-coded status down this is our down
event now we will add a reason so there
is a v parameter you can add to the
passive service service status it's
optional these four you got to have but
the passive reason code we're actually
going to dig up from the trap if you
remember when we did the trap I would
send it in and we had a we'd add a VAR
bind and this was just one dot 306 dot
104 to 1.9
and nine cuz I just know it doesn't
exist and the VAR bind would actually be
something like they're bored or if it's
an up event they got pizza etc so what
this will say is go to this original
moody event which came from the moody
trap take this VAR bond if it exists and
set that as a passive reason code and
then finally remember you always have to
change the UI of the the triggering
event so would have set that to be our
service passive server status event this
is the UI for the passive service status
event that changes the service from up
to down
make sense okay the next section is the
up event this is our joyful trap so the
joyful trap will come in converted into
this event and again we'll do the same
thing we'll set the the node label IP
address etc notice I left off here I
have this reason code I left it off I
mean you could put it in there there's
no real reason but no one really cares
about UPS so no one really I mean if you
do find but usually up means
everything's okay I don't care I want to
know wipe down so I left it off so we
again we hard code localhost 127 I know
no one in the classroom service again
remember that that's important and then
we have the status equals up and then
again we set the event to be the passive
services that passive service status
event make sense okay so now we're going
to update the configuration files again
I suggest you don't type these in you
can cut them paste them from my
screenshots but what I like to do is go
to the training which is again WWF in
MSN org slash capital T training and
then we've got a configuration our
slides of our videos almost an own
config and it's not there because I
forgot to add it okay let's add it
because I screwed up okay I usually
don't do this but I don't like editing
things so here it is live come back set
whoops I better do it as me sure sound
if you do
[Music]
and we'll go to trainings fig directory
so that's ten exercise and I did
actually copy it up there I just forgot
to move it so let's move here okay let's
get out of there and get out of there
okay going back to our screen I'll hit
refresh and magically we now have an
exercise ten directory and there's the
polar configuration and translator
configuration now if you look at the
polar configuration file again don't cut
and paste from here because it adds all
these little bashes and those little
batches over pain-in-the-butt just look
at the source if you want to do that and
there is our class mood service remember
again it's only one line and remember we
have the class mood tag at the bottom to
tell open and that's what class or what
piece of software to use for this
particular thing so what I want to do is
this right mouse click copy link
location go back to my system CD
optimist Etsy we're already there W get
- capital n which will not clobber which
was will clobber the polar can be
existing public configuration file and
then we'll do the same for the
translator configuration
so here you get best captain and set in
and now I'm going to stop up at m/s if
you guys were with me when the last
lesson for some reason our script and
system control doesn't is not happy with
we start and so what I'm basically going
to do is I'm just going to use the
script itself which is a knocked open
and it's been I'm going to stop open at
MS and then once this comes back up I'm
going to start it again the only reason
we need to do this is the fact that we
changed the polar when we did the
original floor exercise we are trying to
move to the point where you do not have
to restart all of open and that's when
the configuration change is made
but we're not quite there yet so while
we wait for that I'm going to go ahead
and look at the translator configuration
file that we have here and as you see
here is our passive status events so
again we're taking this here moody event
which is the trap we're going to assign
these four parameters add an optional
fit parameter for the code set the Yui
as a passive service status for the up
event we just repeat that we leave off
the reason code and instead we still
have to set these four things you have
to have a node label passive IP address
has a service named peso status now we
thought about adding um putting node IDs
in here but we didn't because most
people will probably know even if they
get a trap they can pass something that
can serve as a node label in the case of
the satellite perhaps there's some field
in the trap that comes from the pixel
metrics box that is useful as a node
label so we put that there so you didn't
have to look up the node ID but the
passive status keeper will actually do
that and store that in the in your
database and so we've got that so going
back to our slides
just to repeat again our goal was to use
the passive status keeper to use events
to trigger a service change so we enter
a class mood service to our existing 127
got a load on one interface on our
localhost mode we synchronized the
requisition and then we found out that
the monitoring status was not monitored
because of course we didn't put it into
the monitoring configuration so we added
it single line again there's nothing
really there outset of is it upwards
down and then we added the monitor line
to the bottom now the the heart of this
is the event translator so we had to
take our event and it could be anything
it doesn't have to be a trap it could be
a syslog message it could be an internal
open of the mess event but in this case
we had our trap that we configured in
the trap lesson and we converted that to
an event we set all of the the passive
no label IP address service name and
status in the case of the Moody event to
be down we added the reason code we did
joyful as the up event again their label
IP address serves name and status of up
and then we updated our configuration
and we restarted open in a mess so again
system control restart open a mess it
obviously worked at some point I'm it
didn't work now so I do start and stop
manually now again if you we should be
up by now so let me go back to my system
here and we are yeah we are up now if I
go to my node we should see the class
mood is now monitored as 100% so we
actually have the class mood service
here and it is on represent because it's
up now let's cause an outage so I'm
going to go through my history and look
for send trap so here we go here is a
trap that we sent way back in one of the
earlier lessons and say I for by let's
say yeah let's just use they be bored
they be bored so that's our lesson so
again we're using a trap we're using
sending this to the version1 trap using
a community name of public remember open
an MS ignores three names so we don't
need to care about that we're sending it
to 127
one and we're going to send the the ID
of the trap which we've just made up is
this nine nine nine nine two note this
is very very important because we want
this has to match localhost is the
source of the trap generic type of one
specific type of six this is a blight
placeholder and then we added this
variable binding that is our reason and
also men in there like that now within
thirty seconds we should actually get
vivid well there we go and so here we go
we got the event this is the event that
came in of NMS classes moody they'd be
bored so that's the event and then this
is the passive status event so since we
didn't go and change this one to be do
not persist or logged only we see in the
GUI and this was the passive status
event that came in now at the moment
nothing has happened but by the time I
hit refresh we have an outage so the
class mood status we have an outage and
there's the reason code they be bored
and that's what we sit in in that cool I
just want to stress this is what I love
about this lesson we have an alarm that
the node law service we have an outage
that denote law service we have
notification is what we have notices if
I click on the notice we have this class
mood service so we used an event and we
were able just by sending in a
particular event we were able to monitor
this thing as if it was a service think
of what you've accomplished this is a
kind of cool we created our own custom
event and then we used a trap to
translate that trap into that custom
event then we took that event and we
translated it into a passive service
status event and then we changed the
status of the service I love this
exercise because now everything's
triggering you've got escalations
running you can open trouble tickets
it's just awesome thank well let's fix
it so I'm going to go ahead and send
this trap in because this is the up
event notice it's very very similar
except that it sends in it has a
specific value of two we decided this
cific value of 1 is down specific value
of 2 is enough notice there's mail
that's all of the ups and downs those
are the notifications that we're getting
so that's kind of cool now if I go back
here and refresh this notice the
notification is automatically cleared
because we got the up event there's no
more outages there's still an alarm
because if I look at the alarms this
hasn't kicked in yet notice by the time
i refresh this this class mood should
actually get cleared because of our
generic clear so our generic clear every
30 seconds it's running and it matches
up swift downs and so it'll clear that
event and within 5 minutes all of these
existing events will disappear and so
that's kind of cool so think about it
you started out with doing silly things
like creating a user and putting in a
password I'll say they're silly but
rather simple things and I know some of
some of you watching this kind of
rolling your eyes ok events that's kind
of straightforward etc but by taking all
of these different pieces and putting
them together you can create some really
cool things think about it if you have a
if you have a printer and you want to
monitor toner you can get toner if that
printer sends traps when toter is low
and when todo has been refilled you can
create a toner service if we wanted to I
monitor a drink machine with open in the
mess we could go and use that to
actually monitor I could have is the
drink machine ok and I can have that
service anything on your network that
you can get events for that you can
either create with our send event PL
script or something similar or comes
from traps or comes from sis logs you
can use to really enrich the information
you can provide to your operators and
the way you monitor your network I just
think that's cool anyway
so again outages when we create an
outage we get the outage we get the
outage reason we get a notice we get an
alarm and then when we fix it everything
comes back up we get again the events
the outage and the alarm and that's it
so
that is it for this lesson
I hope you folks have found it useful we
have two more lessons left in the class
so if you've been hanging hanging around
for so long I appreciate it I hope you
find these things useful the next two
lessons we're going to talk about data
collection and we're going to talk about
threshold link
now the fun doesn't stop there I have
big plans for a number of other lessons
but in order to do this there's only
much more fun for me because they'll
just be one-off so if you when you if
you get through all 12 of these lessons
you will have a solid basic working
knowledge of opening a mess
and I hope it isn't as scary as it can
seem just trying to find your way
through the documentation and the wiki
and then you can build on that with the
other exercises that we've got going
anyway hope you're enjoying it please
leave this feedback in the comments and
until next time take care