#networking
1 messages · Page 320 of 1
@rocky badge i'd never let comcast have my phone service
Ubiquity propaganda detected
@flat wagon that wifi thing from telus is a $50 one-off fee
is that a joke...
Ubiquity is pr good
@flat wagon overpriced, violate copyright and also recently had a massive security leak, which they didnt disclose
ubiquity is in the dumpster for me
eh
some products are good
I only have the unifi USW FLEX mini
and the ER-X
from the ISP line
looks like gpon, although idk if the sfp would work if plugged into a mtik or some other device
wow is that my router u found information on
@flat wagon seems like it yeah
damn ur good
I mean doesn't it depends on the module
@peak cloak It doesnt do GPON modules, or rather: only the mikrotik GPON modules
@flat wagon ideally, you'd ring up Telus, and ask if they provide active ethernet or GPON
they can do that
If its Active E, you can skip their gear and plug the SFP+ directly into your own router
If its GPON, you may have to use their equipment
Might be a GPON because there's some fiber box connected to that router (the box kinda looks like a switch but that box is connected to the actual Netgear switch, so idk)
that's just a splice point
o
yes
Yeah its just a wallmounted box
its just a splice point
PON's have 1 fiber medium for every house
you need a special module that can use GPON
ours not wall mounted tho it's on a shelf but the box is by NOKIA and i was surprised they into networking cuz I thought they died off
Active E is a dedicated fiber per house
@flat wagon nokia makes 5G technologies :P
and other networking things
well guess they needed to find income somehow after their phone division went rip
the phone branch of Nokia was sold to microsoft, and has since then died.
yes
@flat wagon but yeah, you should ring up Telus, ask them if they use GPON or not
See this is what my ISP gave me ^
they just gave me a module and a patch fiber
if Telus does Active E as well, then you can (and I would) ditch that telus wifi thing
and use your own
Like this?
That's a splice point
yes
@peak cloak if the ISP modem is "optional" then it is safe to assume it isnt GPON right?
no clue
That's the "splicer"
that's the switch
the oval thing beside the wifi router is the wifi booster
yeah that's not a splicer, it has actual power
what's on the back of that?
The black box is the switch
no what ports does it have on the back?
and thats connected to my router too idk what this is but I think it's UPS backup power
and o
ugh let me see
Honestly no idea I cant get to it
I can't twist it too much the fiber optic cable just reaches it and dont wanna twist it and put strain on it
yeah understandable
where did @tame carbon go
confused by what it's purpose is
same
im here
o
I just was away for a sec
just look at the pics
if it converts GPON to active fiber it would be nice
but you have fiber to the router @flat wagon ?
Okay, and that red cable goes to your router, right?
yes
no
unlike what crystal showed?
@flat wagon that Nokia is your ONT, or Optical Network Terminal
ok cool
I have something similar
it's the big box on the left
on the middle
that's my ONT
yaaaassss another #UBNT fan
and it goes to the router to the right
thy thing is huge
it's the same size of yours probobly
@peak cloak whip it out and compare
lol
Um
@peak cloak I'm very confused now
how?
how the fuck are they delivering 1.5G service
Just do we're clear u talking about the silver no on the left right
over ethernet?
No clue
also could be false marketing lol
what's the model of ONT?
^ yes, give us the model #
no they got a 940MBPS plan and a 1.5G plan and uh ok hold on
don't think yellow is fiber
usually gpon wires are white
that router and optical switch thing seems outdated af
but seems like the same telus ups thing i got tho
what's a P/N
product number
i dont see a model number anywhere
@flat wagon did you recently upgrade to those speeds ? because I'm not entirely sure you even have the right gear to use these speeds
well we were at 940MBPS and we just upgraded yesterday
@flat wagon yeah but that's as fast as gigabit ethernet goes.
and we contacted telus tech support and they said it should be working
@flat wagon okay if you log into your router, can you look at the WAN connection status?
it should list the negotiated speed with the ONT
its either 1G, 2.5G or 10G
Ugh I cant do that atm but ideally it should be 2.5G or 10G?
I can check in an hour
@flat wagon well, I wouldnt want to say ideal. I am just trying to figure out what standards they use
o
If it was a 10G interface, then it would make life easier yes
2.5G would be fine too, would need a module
but if it is 1G, as I am suspecting it is
then you have a problem :P
then what would we replace that optical box with?
a better one
link to a few?
it needs to be replaced by your ISP
gpon isn't standerdized like DOCISS
you can't just buy one
pretty sure that's the best they offer atm
but it's a year old when they were only selling 940MBPS down and uploads
so might be wrong
@flat wagon yeah but is the gear you have, the same
?
Because it is probably just 1G gear
:e
@flat wagon when you get a chance to confirm the link speed to the ONT
tell us :)
because if it is 1G, telus hasn't done their job.
kk
1G wont be enough
@flat wagon see this is what I got ^
the left yellow fiber is from the ISP
but its only 1G
Optical Network Terminal
@flat wagon this thing does what your ONT basically does: https://i.imgur.com/VL8Yvq4.png
Takes fiber, spits out 1G copper
I worked with someone on Discord like 4 months ago but you can ask Telus for an ONT-SFP Module in place of the box and they will give you one
@plain siren can you plug those into mikrotik gear directly?
Its DHCP on the other side
Yes, and Ubi
its the 1.5G fiber service they have
This is something i read online
I assume they give out SFP+'s now?
@flat wagon if what Rouing said is true, then you should be able to ask Telus for an SFP+ module
If you can get your hands on that, you can use your own router and networking gear.
that's sfp underneath?
SFP or SFP+?
calls TELUS
"your estimated wait time is.... 1 hour"
instantly hangs up
@flat wagon you have screwdriver? :D
yes
If you can remove the module and snap a picture.. that would be splendid
I wonder if mine is also like that
yeah I cant tell which model box this is, some have it some dont
doubt it would though
need better pics
@plain siren I had him look for Model# but it didnt have markings he said
Nokia slapped on the top of it?
yes
7368 ISAM G-240G-A perhaps
But that's gigabit only
yep
Nokia 7342 ISAM O-010S-P is the SFP Module we got provisioned according to these old dms
It provides an asymmetric 1.244 Gb/s upstream and 2.488 Gb/s downstream GPON uplink
nice ONT for that module then 
@plain siren Pretty sure mikrotik does not like GPON
says it all over their doc
thing is, with an SFP module you no longer have voice?
not unless you configure SIP properly
okk to be clear that ONT can support the 1.5G down, correct? idc about the upload speed much
or Optik TV, however who the hell really wants to use the Landline service from their ISP in a semi-optimal network config
yeah our tv is connected to the internet and won't work without it
what is this BS then?
the newer SFP-ONT's support 2.5G
Annnnd it turns out you can buy these SFP Modules second hand and do this:
ritool set MfrID ABCD
ritool set G984Serial 012345678
to set the Serial #' to the ONT Box
suddenly connection
on what router?
oh yeahh and
uci set gpon.ploam.nPassword='0x30 0x31 0x32 0x33 0x34 0x35 0x36 0x37 0x38 0x39'
uci commit```
But I can buy my own ONT and it should be fine if I connect everytbing to it?
depends
ONT should be provided by the ISP.
The one you have right now, is not suitable for your internet connection
you SSH into the SFP Module
wut lol, never heard of that
SSH : user ONTUSER, password SUGAR2A041
quite a bit
https://rsaxvc.net/blog/2020/8/15/Nokia_G-010S-A_Pin_6_Issue.html Some ONT's which dont power up right need Pin 6 shorted to ground.
is there SFP onts that work with fios? I've looked online and couldn't find anything
Someone literally ripped apart everything regarding the service feed and its "authentication" for Bell (Telus)
https://www.idscomm.ca/blog/bell-fibe-internet-iptv-with-pfsense < You can get IPTV over the SFP module
VLAN 34 is IPTV
AT&T FiOS?
no verizon
hai melp
sup bud
Yeah, they are alot of fun
whatsup :)
uverse is DSL, att fiber is called att fiber, or did they rename it again?
oh maybe
ATT renames things like every week it seems 
no 😦
bummer
Verizon has some funky shit that we still havent figured out quite yet thats preventing normal OLT<->ONT Spoofing from working. Its speculated they have the ONT's loaded with some extra "fingerprint" that is being served post-initial handshake that we cant quite sniff. Not to mention they are even more picky with their vendor's as they have all 3 big vendors in a mix-mash depending on your local region.
(Side note: https://www.itu.int/rec/T-REC-G.984.3 < ITU does have a standards document for GPON Networks)
yeah GPON is standardized it's just all stuff that's annoying
[Side Note Disclaimer: Its "Recommendations" and meant more as a workgroup focus point]
yeah
Optimum also started ftth here
Optimum is basically altice
It's even worse because they don't provide an ONT apparently
just direct fiber to all-in-one
the ONT is on Verizons Side
and apparently from reddit posts passthough doesn't work
Thats the big oof
gpon would be fine
we have to come up with shitty standards
if it was just standerdized
so that people need to use confusing hardware
kinda like dociss
@peak cloak GPON feels more like a cost saving measure to me
no
Yeah you are telling me
@peak cloak if my fiber was wasted. they can just blow a new fiber through the tube
with PON you have to re-do every splice/tap
AT&T is leading the industry development standards for Residential PON Networks
So thats great
not everywhere
plus ftth is on poles
if my fiber is cut then they just tap a new one
@tame carbon US lol
it's everywhere
cost saving
it is
every year you guys have hurricanes
and then some traffic light falls over
takes out a power pole and transformer
like, what the fuck
omg.
yeah and?
Profit today much better than profit tomorrow
underground isn't practical here
Oh yeah and heres the packet walkthrough of Residential Gateway <> Virtual Subscriber Gateway. This is basically all the shit on top of your internet
@peak cloak YOu know what is cool? Brick wins vs Wood
when i lived in philly, it was all underground
If a tree falls over here, and hits a transformer, nothing would happen
and it was fantastic, hardly any power outages
in cities it's underground
because it is completely encased in a concrete house
Everything where I am is Underground by thats prob because... looks and the whole HOA + White Oil People and their Backyards
@peak cloak the local fiber splice box that takes the big 144 bundle and splits it up to 2 per house
thats under the ground, by like 30cm of soil
when you remove the topsoil, you have a hatch
and they open that, you can remove the splice point
it takes a crew of 2 about 4 hours to completely redo a fiber run to a house.
this is what good permanent infrastructure should be like.
low maintenance
if only everything was implemented with that kind of foresight
lol when they were setting up my fiber service it was just one guy and a ladder
@peak cloak the day they wired up the neighborhood, they sent 1 literate guy + 8 syrian refugee workers over
they operated the spades and rocket drill
they did 12 houses
in 1 day
Our has man holes that can fit 2 people for each splice box, UG Switching System, etc. Someone got mad apparently they couldnt see and there is strung up LED lights yoinked around the hatch lip with an 18650 battery pack.
It was kinda funny to see
@peak cloak There's three teams.
The guys who do the physical pipes.
THen the guys who run the fiber to the local node and install the patch boxes
and finally
they call up customers to make appointments
to blow the fiber through the tube and install the indoors
After the splice to the houses has been completed the day after
Easement Conduit Trenchers
Specific Metro/Municipal Infra Contractors
Residential Install Contractors
Always
I sent angry email
and suddenly one spawned in front of the door the next morning
after months of failed phonecalls
The cities that are smart have the Municipal/Metropolitan Infra Installers all show up at once and they can run Eletric, Gas, Debug Cable, and other shit all at once
But you cant do shit for residential
Usually if you get Comcast, AT&T, and lets say Verizon in the same place, they will actually work together and get shit done faster. Its amazing
you see AT&T on one side feeding line and comcast on the other pulling pull ribbon
If I had a network with 2 switches with each on a different subnet and 2 VLANs on one of them would devices one VLAN be able to share IPs with devices on the other VLAN without colliding?
umm, why share IPs
you can't
also that's not enough info
I mean technically you can, but that depends on the router config
are they being NATed?
or just routed?
your VLANs need to be on different network space, so no, using 2 VLANs won't let you share IPs
unless you're doing some other stuff on top of that
like NAT
you can't (for example) have two VLANs, both on 192.168.0.0/24
you'd have one on 192.168.0.0/24 and one on 192.168.1.0/24
also many people confuse VLANs with subnets
VLANs just simply carry subnets
but they can do other things too
You can IF they aren't routed anywhere, ie the 192.168.0.0/24 subnet doesn't have a gateway out
Obviously there isn't much point but if you want a local network that doesn't need access to anything but the local subnet you can overlap networks
right, but i doubt that's what he's doing
Yah, more information/clarity needed 🙂
what are you trying to do exactly?
@tame carbon warship?
@peak cloak for your collection: https://i.imgur.com/GJVIhCR.png
no, that was jsut copypaste fail
I speak with some authority though
is that evading the filter???!?
im gonna call linus
The switches can be on separate subnets themselves. there would have to be a root switch that would trunk the VLANs down to the second switch. In which case Devices on the Root switch with a matching VLAN the same as devices on the secondary switch (with the same VLAN and Subnetting) could talk to each other.
the acronym for the popular "network UPS tools" software triggers the filter as well
@untold elbow come again?
for some reason he wants to have IPs to not collide?
would devices one VLAN be able to share IPs with devices on the other VLAN without colliding
I dont think he knew what he was asking, but judging by the "intent" of each acronym,
Could you use the same IP Space across 2 devices on the same VLAN on 2 diff Switches? seemed to be the question intended
exactly
yikes
@untold elbow I bet louis rossman would get muted permanently
when he starts talking about his **busses on the boards
I forgot I had gotten angry over that censor a few days ago
I was trying to say.. Nt and bolt
bolt?
Yeah
what if you make it plural?
nuts
oh hay
Mm?
reported for evading the censor
Litmus on his way!
YoUr ReGEx Is TrASh aNd YOu shOULd FeeL baD
wait... does that really trigger it
Yep
what
right?
Someone kept saying it ironically and ruined it I bet
I formated my laptop now my RJ45 connector working perfect just the lights burned out
might be a drivers thing
I don't mind the lights if I have internet before the format the rj45 connector didn't worked
It's an old laptop with little improvements
6 years I have it or 7 idr
Never touch something that work
I learned it the hard way
Finishing the LAST assignment to get my degree
its something from years ago
but its fucking BORING
I find it awesome AT&T puts all their infra framework designs out there for public view
Feels like I am writing usecases for a stupid candy dispenser
Markdown
is
kinda
neat
i can never remember the hyperlink syntax
the bbcode sytax is way easier
BEEP
ew
bbcode 4 lyfe
trying to get work done over here ok
@untold elbow use pandoc.
When I had to write documentation for mediawiki
that syntax is even worse
@untold elbow yeah what you think this is ? :D
I inject some \tableofcontents in there
and it parses the markdown # headings
and generates nice document for me
Pandoc is great
Oh holy shit the AT&T ONT is SSH able. Excusemewhat
write markdown, convert to tex, and generate pdf
you can put tex in ahead of time, and it will just stay untouched
and even add headerfiles and import packages and such
whats that mean
can i get free tv?
Even if AT&T whatever TV they have now name here was accessible, still wouldnt be worth it
thats snazzy-looking
does that render the markdown live?
cool
as you go through source
what's the IDE?
IntelliJ
oh ok
neat, i like it
If you have ultimate, it does all languages
i don't really have to write any markdown ever, but good to know
DataGrip is fucking awesome
ugh
It does it all
as long as i dont have to npm some stuff, im ok
i wish sublime text had something like that
for snippets
you use yarn?
@untold elbow its like npm
ah
but different
oh yeah
yarn is a task runner
I just want to use maven over here
and intellij integrates with it fully
Oh how nice, I found the 802.1P priorities used by AT&T
so IDE build uses maven
any software dev stuff i do is just dumb hobby stuff on my own, so i get to be picky
picky and lazy
I got sunk so deep into this now
that I am so used to my tools now
lol the amount of plugins bolted onto my IDE
I have... complete devkit for android, PHP, Python, NodeJS, Java and C++
it has this so you don't need huge node_modules
SQL database integration, some Parser generator tools
An overview of Plug'n'Play, a powerful and innovative installation strategy for Node.
and it's just global
@untold elbow oh ye and they added a Cypher plugin, so I can now do Graph databases straight from source too
Autocomplete on queries and even have a built in graph viewer
_>
only server-side stuff i've ever done was for a simple disk monitoring UI for my NAS: http://h.jro.io:58765
and that was with flask + socket.io
i dont know what im doing
lmao
im not a web dev
socket io lol
looks like it's workin great to me 😉
yeah but this isnt how you do sockets xD
shouldn't it be one socket?
no idea
it's connected to a few systems in the background
instead of one single thing your updating
😄
how do i fix that
@untold elbow server has to use keep-alive, or instead use a websocket
have one single object to transfer instead of multiple small ones
websockets is just a TCP socket that stays open for the duration of that session
what you are doing right now, is just opening HTTP connection, asking for a file, then closing connection
https://cdatatec.com/product-item/1u-16pon-port-pizza-box-epon-olt-fd1216s/
Pizzabox OLT
What the hell is a pizza box?
dumb name for a 1U server
Wooooooow
socket.io reverts to http long polling it says if sockets aren't available?
@untold elbow benefit of a socket is that the server pushes, instead of having the client poll
welp maybe if i get really bored i'll fix it
"Maybe into some other box"
How descriptive
@peak cloak yeah but longpolling doesnt work in this context
since you just want to stream
again, im not a dev
so dont judge
you can tell socket.io client to upgrade to websockets
That would be
imma look into this
if it was nodejs.
and keep alive
Because this is python
Once you have installed the Socket.IO client library, you can now init the client. The complete list of options can be found below. In the examples below, the io object comes either from: the
it happens on the client side
server has to decide on keep-alive
because it sends the headers
Client can only suggest it I think
yeah
or this too
Once you have installed the Socket.IO client library, you can now init the client. The complete list of options can be found below. In the examples below, the io object comes either from: the
@untold elbow strange. Very confused by this code
you emit to.. redis:// ?
wat
socketio = SocketIO(app, message_queue="redis://")
what is this lol
Not seeing you spawn an HTTP server anywhere
must be that library
It looks like we may be able to steal neighbors services/TV
# Start flask/socket.io app
@app.route('/')
def index():
return render_template('index.html')
if __name__ == '__main__':
socketio.run(app, host='0.0.0.0', debug=False)
so you are using flask?
Also, it also may be possible to actually virtualize the ONT. In docker. They have a vONT Docker Container
@untold elbow Oh I see your problem.
that's socketio
@untold elbow before you run app.start() you should put this line of code: WSGIRequestHandler.protocol_version = "HTTP/1.1"
That will set the webserver to use HTTP1.1 because its using 1.0 right now
then you should have keep-alives
from werkzeug.serving import WSGIRequestHandler
yeah, I tried manually changing the client code to use websockets
didn't work
so server problem
ah
@peak cloak HTTP/2 is even better
though debateable
it supports binary transfer
request pipelining is gonna be a big one moving forward
because then you can multiplex :)
1 connection all your requests, then you close.
Basically fixing the Apache 10K Connection Problem.... the correct way
or use nginx
Instead of the NGINX Async software layer method
yeah
nginx will scream with this
@plain siren it might make a lot of sense for load balancers
Load balancer <--> app can be a single connection
much better threading options and fine tuning
Traefik is the first thing that comes to mind
AWS AppMesh is another thing that also comes to mind
@plain siren I've only done load balancing in lab contexts
we had built a couple completely stateless microservices
and had a common runtime engine that could spawn up these instances and bind them to a proxy server
so you could run backend auth payment tracking to spawn 3 services
and all the services that were running within the same instance could use interface proxies
if the service it needed was not in the local context, it would contact a load balancer
the entire software stack was dynamic proxies
that were either direct interfaces or thin REST layers
Yall rebuilt k8s network control plane
using some clever dependency injection tricks :D
@plain siren this was for a route sharing service
that had to be scalable across many universities
Route sharing?
err
ride sharing*
it hooked into the schedule system
and used google maps to calculate trip times
all the routes were cached, and we used dijkstra's algorithm (with some heristics) to allow people who live along eachotherrs routes, to be added to the ride
Something like this
Something like that yes
ofc the ALB can point to many autoscaling groups
except we used redis only
@plain siren this taught me a lot about writing stateless services though
sessions are the most godawful thing ever.
yes
@flat wagon how'd the call go?
so im using redis to allow communication between threads... i forget exactly what i was communicating, i built this a few years ago and wrestled with it for months, barely got it working, then gave up on it and haven't really looked back since
and that http1.1 thing didnt change the keep-alive. im going to fix it by removing the NAT rule on my network so you cant get to the page any more 😉
nope lol
this is only accessed by me when i want to check the disk temps
that's my personal NAS system
@untold elbow are you in your router settings rn?
doing some heuristics on your latency https://i.imgur.com/s62ukIQ.png
lol
you stopped talking
yeah i was on a call
and response times got all wonky
oh i restarted the app w/ those changes you said
@untold elbow oh so its using 1.1 now, but not keep-alive mh.
honestly that socketio is just strange
yeah i think i did it right
but the rest is pretty cool
but all of this is just stuff copy/pasted from stack exchange
@untold elbow you want an idea for v2 ?
you install influxdb, and write your values there
pretty easy actually
api is json
yeah i need to do that
can i build something similar to the table display i have now?
@untold elbow https://play.grafana.org/
my issues with a lot of those dashboards is they present stuff in a way that isn't super useful to me
doesn't influx have a visualization thing now as well?
yeah but grana is better
@untold elbow all the queries you see are on the same timeframe you have at the top.
looks neat
you can basically zoom in at any level
you can do refresh yes
I just randomly read this one line by itself and I thought you were saying your grandmother is better
but ~5s is stretching it
yeah on the top right you can change refresh rate
@untold elbow the time frame system it has in the UI is bound to all your database queries as parameters
you pass those into your timeseries query
and you can automatically get series of values for multiple lines on a graph
you could configure it like, a unit of server
and a server has multiple measure points
you can set up combo views for all of that
and when new points come online
they get added to the graph automatically
so its very flexible
i like it
you can go into editor mode
you can either enter a query by hand
or use the builder
and then you can apply functions https://i.imgur.com/BQJ45hf.png
or do fourrier transform xD
this is all part of influxdb btw
def a project for the future
limited free time right now, have a 6 month old at home now
and just dumped all its registers every 10 seconds
soldered an ftdi chip onto a cable and reversed the polarities according to spec
wrote tiny python program to read from serial on a raspberry pi
that was just sending the data to my influx instance on my server
and then just come grafana tinkering to get some cool results
data querying ^ is pretty simple
mmmmm Grafana 
When I went into a Zabbix chat to try to figure out how to get the UI to be more user friendly, they recommended Grafana. Then I had a look at what Grafana could do, but I don't think it can do what I want
Grafana can do all the things
You can set up event hooks in grafana
I use Zabbix for SMNP and Grafana to make it all pretty
Grafana takes alot of time to configure is the real crutch
^
where we can enter a device and see the graphs, latest data, and problems for a device all in one place
yeah you can do all that, any index can be used on any axis of the graphs or tables
And there are plenty of them
https://www.fluentd.org/ Combined with this guy its a lovely mix.
yes I see that screenshot of different types of things you can create
but we have like 300 devices that would need that per-device view
Thats pretty easy
like do I have to build this per device?
for each one individually?
that's what it looked like when I was looking at grafana
you can template it
with influx stuff is quite easy
Or pragmatically deploy the configs
you can just attach properties to each batch you send
and then filter by web01 web02 etc
and have the same points for each of them: cpu, memory, network
pragmatically? do you mean programmatically?
Yes, but it annoys people
right but that's four web servers in one graph
let me show you what sort of thing I am trying to get
You can get a page per device kinda view going too
For example, take a look at this random google image, look at the top where it says "host"
You can have tons of hosts listed there with the same dashboard layout for each
we can go to events to view the events, go to graphs to see the graphs
ok
I need the hosts to be grouped in a sort of tree structure though
as opposed to one big flat list
they also said I could do that host tree in grafana
I don't want a big drop down list with all the hosts, it would be a bit crazy
I also don't really want to have to design the graph page myself, I want grafana to automatically pull every single graph from zabbix that there is for that device
and provide some means of changing which graph you are seeing
@tender hazel you can also just use grafana as add-in
you can generate embeds
they are dynamic too
let me show another thing
in zenoss we have this tree view of locations and sublocations
we can navigate through the tree and click on a location
I would want to replicate that in grafana somehow
the zabbix guys said grafana can do it but I haven't seen any examples like that
so something like this perhaps
Ok I havent taken a good look at the indicator features of this personally and this is cool
I gotta go mess with this myself heh
all of the examples that I see are sortof overall dashboards, which isn't what I want, I just want an easy way of navigating between devices and seeing everything for a device
This is what you are looking for: https://grafana.com/docs/grafana/latest/variables/
It allows you to have a single dashboard and use variable to apply to the entire dashboard. Essential will let you do what you need of type in the node name and it will change all the graphs
Templates and variables A variable is a placeholder for a value. You can use variables in metric queries and in panel titles. So when you change the value, …
@hollow marlin oh yeah, those cascade
so if you change the left value
the other choices change
ok so that would work for changing between graphs for a single device
what about the tree of devices?
for navigating to a particular device?
Yep, I was looking into it for a presentation for a dashboard for our NOC. I have not tested it but it would be what you Fin is looking for
that you can navigate through and click on a device to view the page for that device
Oh are you looking for a network map?
@tender hazel Ever heard of Icinga?
This is more of a ... enterprise solution buuuut
yes some kind of a tree structure like that
I want it to be automatic
we have the tree structure sorta set up in zabbix with groups
we replicated our tree structure from zenoss
so we have groups and subgroups
but there is no option to navigate through the list of groups and subgroups like there is in zenoss
yeah I don't want to have to create dashboards folders
I want for grafana to pull the tree structure directly from the zabbix database and organize the devices in that way
zabbix has a "map tree" built in that does something like what I want
the issue is that viewing a map is not always the most efficient way to browse to a device
some of our larger sites have like 30 devices so the map has 30 items on it
when you are trying to find a specific device you have to look all over the map to find it
when it is a list instead of a map, it is sorted alphabetically
so it becomes rather quick to find a specific device
@tender hazel really sounds like something custom you'd want here lol
Rendering a tree view isnt that complicated with a bit of code
if you can just populate the right side of the panel with a grafana view
you dont have to use their dashboards.
oldschool frameset
you click on an item on the left
and the right side updates the url
yeah something like that even would work
I just have to make it more user friendly because we have level 1 service desk techs using it
and when they have to look at a device and see what is wrong, I don't want them to have to go to latest data->browse for device->apply, then look at latest data, then go to a different part of zabbix where graphs are, and browse for th device again->apply, then go to a third part of zabbix problems->browse for device-> apply to see the problems
it is too much work going to diferent parts of the system to find everything about a single device
yeah you want to just have small queries
that fill out the info that is relevant
when you click on a page in the treeview
yeah those sort of tree views are what we want.. and as I said zabbix has a map tree that is sorta like that but clicking on something in the tree calls up a map instead of a list of devices
if you want something user friendly and specific
you are looking at custom solutions
I went to the zabbix chat to suggest to the developers that they make a device tree view but they basically said "no, everybody loves our UI, it is perfect, we don't want to change anything"
Yus
https://material-ui.com/components/tree-view/ It uses this UI
It goes on to make further logic breakouts in another "Advanced" blog post about this
https://technology.amis.nl/frontend/further-customizing-a-grafana-custom-plugin-in-react/
@tender hazel very common btw for custom management dashboards: https://adminlte.io/
AHhhhhhhh yes the most copypasta'd UI for WebUI's
yeah I don't want to have to use adminlte.. the react tree component looks promising
what's the back end?
rest service
the page controller is just MVC
written in java
but the data is fed by a rest api
that's another java process
@tender hazel oh no I reinvented the wheel on this one
wrote my own MVC page router
and controller structure
its incredibly simple
@tender hazel those two views you just saw:
@Path("status")
public class StatusController {
@GET
@Path("service")
@RequiresLogin
public ModelAndView getServiceStatus(Request request, Response response) {
Map<String, Object> models = new HashMap<>();
ServiceAPIClient client = session(request).getClient();
models.put("services", client.getStatus().getService());
models.put("servers", client.getServers().getServer());
models.put("page", "Status");
models.put("page_desc", "Services");
return new ModelAndView(models, "status/service");
}
@GET
@Path("minecraft")
@RequiresLogin
public ModelAndView getMinecraftServerStatus(Request request, Response response) {
Map<String, Object> models = new HashMap<>();
ServiceAPIClient client = session(request).getClient();
ServerInfo info = client.getServerInfo(request.queryParams("server"));
models.put("server", info);
models.put("page", "Server Information");
models.put("page_desc", info.getName());
return new ModelAndView(models, "status/minecraft");
}
}
its just a really thin layer
actual data it passes over is just from the API
yeah I don't want to have to use java
xD
@tender hazel simple nodejs program would do in this situation
java isnt ideal to write webapps with
its more for services
callbacks give me a headache.. I prefer working with laravel/php
I think for this I don't really need to have the async capability of nodejs
which I know is very powerful for multithreading etc
but I find callbacks to be ungainly
I would have to learn about promises
its actually not that bad
its just the language that sucks.
in typescript, writing async code is a breeze
@tender hazel if you put the word async in front of your methods, it automatically wraps the return value as a promise
so inside such a method, you can call upon async functions without having to worry
if you wish to evaluate (block) the promise, you can do await method()
code such as
res.locals.umb = await this.client(this.prepareSession(req).culture)
is a blocking function
I just worry about going down the rabbit hole too
recreating an entire new UI for zabbix
It is a rabbit hole
because once you create it
you have to maintain it
and stuff always breaks.
somebody already made something like what I want once
but it has vanished
the website has been down ever since I found it at first
it's a shame because it looks like exactly what I would want
@tender hazel https://www.ricca-it.com/it/
Realizzazione infrastrutture IT, Datacenter, Sicurezza Informatica, Monitoraggio IT, Software e Web Applications.
Found it
@tender hazel
:D
I didn't click on eCare because it sounded like some kind of customer service software
Visualizzazione
@tender hazel check this out tho: https://snipeitapp.com/demo
that's more of an inventory system than a monitoring system though
Well to be honest, most implementations of what you want are a combo of ITAM, CMDB, and Monitoring
yeah.. zenoss is designed to be a combination of CMDB and monitoring
Most of the time I see these integrated separately but they usually have cross-integrations that works out nice
zabbix sorta has an inventory system but there is no sortof "device view"
https://www.i-doit.com/i-doit/ This is the biggest "mix mash" complexity I have seen personally
Möchten Sie endlich zu einer funktionierenden IT-Dokumentation gelangen und selbst entscheiden, wie Sie dokumentieren? Entdecken Sie i-doit.
they assume you have a sort of workflow where you are interested in all graphs, so you go into graph mode and then choose which device you want to see graphs for and choose what graph you want to see
and then maybe another time you are interested in all problems and so you go into problems mode and choose which device you want to see problems for
but they are lacking a solution for if you care about everything that is happening with one device
interesting..
our server admin has made the choice of zabbix though because it is easier for him to administrate and is the most popular
and he's already spent a great deal of time setting it up
librenms has this sort of device interface too which is really nice
I like using librenms for smaller deployments, but I think we have too many devices for it
