Is this worth pursuing? (PING gadget)

[SL4P]

To fill in some time, I put this (ICMP ping) gadget together...

Typical applications monitoring :-

- Multiple IP cameras,
- Servers up
- Desktops 'on'
- ISP network outages etc.
- Long haul links
- Gaming servers

Currently up to 8 user-definable target hosts, could add more, but blocking delay when 'none' are responding... is quite tedious!

Detects 'no reply', or 'slow' reply... adjustable warning thresholds and reporting detail as per menu (attached image).
Large shot shows startup and 'Detail=2', smaller shot shows 'Detail=1'
About the size of a deck of cards, with an RGB ping-pong ball indicator on top.

Green = all good,
Yellow = not quite right,
Red = hard or persistent problem encountered.

LED notification holds for the duration of the outage,
-- SD card logging coming -- log is overwritten to capacity of the card.

USB serial live monitoring and configuration.
Maybe add a POE option if desirable.

Is it worth pursuing as a product to sell ?
What price point?
--- PM or post as you prefer ---

Cheers

[Jebnor]

Long story short: No.

Anyone who has a number of things to monitor probably has a real server and will whip up a script to ping these or will have a proper montioring service running.

[SL4P]

Fair enough.

I made it for myself from parts on hand in spare time
- to monitor ISP outages, and IP camera connectivity.

The other potential applications just sort of came to mind.
I thought it may be a cottage product for less than the cost of writing a script!

I reckon low volume (hand made) in a nice plastic box, it costs me about $20 to make.
That's about 20 minutes of desk time for a salaried dev/sysadmin.

Oh well.

M

[Alana]

Back in the days when amateur ISP's were common in Poland we had use for such devices but more as automatic restarters. No ping for 2-5 minutes, it power cycles the device it monitors, usually a WiFi transponder.
I think such thing could still be useful if possible to be made cheaper than Raspberry Pi that can be hooked for the same use.

[kolbep]

When I worked for an ISP, we ran Nagios Network Monitoring on 2 of our servers to oversee the infrastructure.
This system ran a bunch of scripts, etc Ping, Server Load, Disk Space, Line Usages, Signal to Noise ratios, Email and Webserver response, etc
It then Emailed, and sent an SMS using an SMS gateway when things went wrong.
The autoblocking was also handy, so we just got 1 message saying the clients server is unreachable, instead of a SMS for every single service that is unreachable.

We also used MRTG as a graphical logging system for almost every variable we could find.

Mrtg & Nagios should be able to be installed on a RPI, as they are mainly a collection of Linux scripts.

[SL4P]

Thanks for  the comments.

I've used a few different network monitoring tools in work environments as well...  sadly these require licensing, a box to run on (the cheaper SW-only options), and someone knowledgeable to set it up.  SNMP is the right way to go if you have all the resources available!

This gizmo is being used as that ''cheap, no fuss widget that simply works'' when you want it.
Virtually zero resources needed - $50, a plug-pack and a network port!

Maybe useful for those punters with flakey ADSL or cable services, and gamers that are watching congestion and latency.
All criticism and suggestions welcome - no trolls please !

[made2hack]

Are you planning on building it to sell?

Or will you be sharing the project with us? IE Open source?

[orion242]

I just use a rasp PI with a small python script and a few of the GPIOs.  It constantly pings a few things and if they fail long enough, they get a reboot and I get an email.

Use these guys to ping things from the outside. http://www.websitepulse.com/  Only runs a few bucks a month to get pinged every 5min from just about anywhere on the planet.

I would have purchased something to do this, but the price would have to been in line with a pi, cheap power supply and a bit of glue for the switching outputs.  Could be a nice niche product.

[SL4P]

... the price would have to been in line with a pi, cheap power supply and a bit of glue for the switching outputs.  Could be a nice niche product.

That's what I was thinking.
Now all planned features are working.
-- Running on AVR/MEGA w/42KB code (can probably shrink it a bit)
-- W5100, (could change for ENCj to reduce cost)
-- NTP time stamping, SD logging etc all good.
-- Seems to be quite error/fault tolerant of most events.

[orion242]

Well post back if you decide to sell them, I may want one.

Another thing to think about is allowing multiple IPs on a single output.  I found rebooting my modem when a single site went down problematic.  That site goes offline and your in a reboot loop.  I now use 3 sites from different corners of the net, if they all fail a few times, then it gets a kick in the shorts.
 

[SL4P]

RENAMED THREAD:  PING hardware gadget

Not sure what you mean "allowing multiple IPs on a single output"

- Do you mean the gadget has multiple static IP addresses ?  If yes, then... maybe... but the default (no config) is DHCP for the device, Static IP is configurable and saved to EEPROM for next reset, but then I need to look at brick mode recovery options!.  Probably not highest on my list. 
(One conversation here for AVR/Atmel multiple IPs http://forum.arduino.cc/index.php?topic=61334.0;wap2)

Is it worth supporting more than eight targets? 
If the network or multiple targets are not responding, it can take a while to complete a pass...

I'll consider anything that works in the current hardware without pushing the cost or reliability out.
Currently looking at  ---

-- a beeper for persistent 'red LED' errors, sustained for 'nn' contiguous samples - and maybe a relay output to match.
-- one LED per target (a row of LEDs - all green = all good) - gimmick value, you can watch the ping phase as it progresses!
-- save/restore running configuration to the SD card (for cloning multiple units)

I'll let you know if/when I put together a couple of 'production' quality

     EDIT:  Adding telnet remote access to the command line interface.
     Works fine, but has an issue when remote client re-connects...

[SL4P]

Just to add...  NEW FEATURE!
I decided against adding a web-server for now, but you can telnet into the unit - to see exactly the same CLI as the USB serial session.

So local or remote management of the configuration is simple.

[SL4P]

Back in the days when amateur ISP's were common in Poland we had use for such devices but more as automatic restarters. No ping for 2-5 minutes, it power cycles the device it monitors, usually a WiFi transponder.
I think such thing could still be useful if possible to be made cheaper than Raspberry Pi that can be hooked for the same use.

I actually made a bunch of those about five years ago.
2 ports in a box, or 8 ports in a 1RU case, under independent control / power sequencing.
An optional heartbeat via serial/USB.  If the heartbeat disappears, the unit powers down (output 0) for 10 seconds, then fires it up again.
They were used on unmanned mountain tops to restart 'locked up' MPEG encoders & decoders feeding TV transmitters!

[orion242]

RENAMED THREAD:  PING hardware gadget

-- a beeper for persistent 'red LED' errors, sustained for 'nn' contiguous samples - and maybe a relay output to match.

Does this have relay outputs for power cycling the devices its monitoring?

That's what I was looking for before rolling my own.  Something that could automatically reset residential cable / dsl modems, PC, etc when a connection issue arises.  Think of remote sites (vacation home) and the stupid dsl modem locks up killing your IP cams.  Home server / cameras and your traveling.

my 2 cents

I don't pay much attention to the logs anymore.  I get the emails (log attached) when something fails, and few minutes later another email(s) that everything got rebooted and is back up.

Even a single relay that you could set to one or more of the 8 target IPs would widen the market for this device a good deal I suspect.

[SL4P]

Yep - a good idea.
The relay was coming anyway - bit now a clear purpose for it !
Thanks

[SL4P]

UPDATE:  Core development is virtually complete.  Packaging / Pricing etc TBA

All functionality is up & running.  USB serial, Telnet interfaces all done.
Maskable alarm relay to do whatever you want - upstream alarms - or power cycling!
WEB interface for dummies will be finished i the next day or two.  (I had to upgrade the CPU for more RAM)

In fact - with the event logging over time, I've identified & replaced a wireless access point that reboots itself every 15 minutes... so the idea is valid!

Crowdfunding next.  Contribute to my market survey at
https://www.surveymonkey.com/s/WS3PHQV

[David_AVD]

On a related note, I made a USB controlled relay that power cycles my ADSL modem in the event of a connectivity loss.  The Linux box that sits on the network runs a script every 15 mins to ping Google a few times and activate the relay for a few seconds if all pings fail.

[SL4P]

Code and design is 99% - phew!

Web UI is now working - a lot easier to navigate than CLI but the nuts & bolts are still down there!   Important to hide some actions from dummies.

Config settings save/load from SD card, along with logging of records & events.
Web UI file download of filtered log records.

Relay configurable to act on specific failures - e.g. to power-cycle modem etc,

75K of code  13K of RAM  188 bytes of EEPROM
Still very responsive on web page!

What sort of $retail would you see this at?
Device, plugpack & manual on disk - a bit like buying a modem/router.

JPG shows about 25% of the web UI page...

[MrsR]

Cost of components can make or break your idea.
eg. last year we designed a board and the only reason it went through was that we got the components cheap enough 1000 ATMEGA48VI FOR $00.20 AU ea. etc.
Dave did a video of him chasing up parts which in our part took longer than having a 1000 boards made.
Good luck with it.

Rachael

[SL4P]

Thanks.  I've made some of those mistakes before !
The ATMEGA I'm using is unfortunately expensive in small volume (abt $USD6-7 each).  Everything else is a known quantity.

Otherwise I'm currently seeing a cost to ship of about $20 each in quantity 10 order. Including enclosure, packaging and etc.  Hoping to RRP for $70-99 each, with actual sell price in the $60-80 range.

All surface mount, very small footprint - could OEM to others for their own integration.

[daqq]

Nice project! Reminds me of: http://thedailywtf.com/articles/ITAPPMONROBOT

Good luck with the project, seems useful!

[CaptCrash]

I like the idea of the device for remote locations that send data home.

My suggestions for change would be as follows
note: Im guessing you may have come up with or discarded some of these options as feature creep, but Ill list them anyway.

1. Increase the number of devices that can be polled so something significant like 32 or 64.
This way the device can be targeted at home and small business users

2. Allow independent delay, latency, attempts and failure thresholds per device with a system wide default setting for simplicity.
eg. Pinging a VPN link, may loose 2-3 pings whilst it is being established.  But then loose none until a link timeout occurs.
You would want to change the latency and failure thresholds and probably the retry vales.

3. Change the configuration options so that your non-DHCP and DHCP ip ranges must be in the same IP scope (or warn against this).
eg. Power outage, both the Ping unit is powered on at the same time as the router.  Router takes longer to boot.  Ping device uses non-DHCP ip address.  Router starts.  Ping device power cycles it because it cannot ping the network.
Also it would not be accessible on the network during this process, requiring someone to power cycle the ping device once the router was up and running.

4. I would add in the option of supporting multiple external hosts for events like a reboot/power cycle of a connected device.

eg. if you monitor a single server (work email server from home for example).
And that server gets rebooted (taken off line for the day for a restore/upgrade etc).
The device will consider the link down and reboot the gateway router and probably keep doing so until the single server is available again.

I think this is what orion242 meant in his earlier post not multiple IP addresses for the Ping Gadget.

You could allow for a number of critical devices (eg. ISP DNS servers, ISP inbound gateway).
Other items, further upstream than the ISP link would be silly to reboot against (and have nothing to do with the local routers ability to manipulate the availability of).
Or for critical devices you could traceroute them and take action when the hop count is reduced.  every 10 minutes when pings are successful, trace route to record the number of hops.  Then on a fail, you can take action.
eg. Ping device -> ping failure.  Trace route  device -> hop count less than previous target -> reboot

In terms of a price point, I would have thought that you would need to keep it under the cost of a home ADSL type router to attract a decent number of customers.  This would mean sub $50, which would be a bit more difficult.

hope that is all constructive

[SL4P]

I like the idea of the device for remote locations that send data home.

My suggestions for change would be as follows
note: Im guessing you may have come up with or discarded some of these options as feature creep, but Ill list them anyway.

1. Increase the number of devices that can be polled so something significant like 32 or 64.
This way the device can be targeted at home and small business users

Not a problem, but with current code, a non-responding target 'blocks' - so the device can get very slow in a storm of failures.  Looking into a non-blocking rewrite.  The actual code is there to support many more (probably 32 total - EEPROM limited)

2. Allow independent delay, latency, attempts and failure thresholds per device with a system wide default setting for simplicity.
eg. Pinging a VPN link, may loose 2-3 pings whilst it is being established.  But then loose none until a link timeout occurs.
You would want to change the latency and failure thresholds and probably the retry vales.

Already in there.

3. Change the configuration options so that your non-DHCP and DHCP ip ranges must be in the same IP scope (or warn against this).
Interesting thought, but I also know of corporate networks where this may conflict.

eg. Power outage, both the Ping unit is powered on at the same time as the router.  Router takes longer to boot.  Ping device uses non-DHCP ip address.  Router starts.  Ping device power cycles it because it cannot ping the network.
Also it would not be accessible on the network during this process, requiring someone to power cycle the ping device once the router was up and running.
I'll certainly think about this.  Perhaps an optional force DHCP mode to wait/retry until DCHP is assigned..  But in most business environments they'll want it static anyway so it can be NATted etc.

4. I would add in the option of supporting multiple external hosts for events like a reboot/power cycle of a connected device.

For cost and simplicity it only has a single relay, but configurable to which target failures can fire it.

eg. if you monitor a single server (work email server from home for example).
And that server gets rebooted (taken off line for the day for a restore/upgrade etc).
The device will consider the link down and reboot the gateway router and probably keep doing so until the single server is available again.

Yes - so you either choose a different host, or set the retry & thresholds to act after ... say 10 mins of outage.

I think this is what orion242 meant in his earlier post not multiple IP addresses for the Ping Gadget.
ahh ok.  Makes sense now.

You could allow for a number of critical devices (eg. ISP DNS servers, ISP inbound gateway).
Other items, further upstream than the ISP link would be silly to reboot against (and have nothing to do with the local routers ability to manipulate the availability of).

Already in there.

Or for critical devices you could traceroute them and take action when the hop count is reduced.  every 10 minutes when pings are successful, trace route to record the number of hops.  Then on a fail, you can take action.
eg. Ping device -> ping failure.  Trace route  device -> hop count less than previous target -> reboot

Maybe v2.   It could be intimidating to setup for small office / home user.

In terms of a price point, I would have thought that you would need to keep it under the cost of a home ADSL type router to attract a decent number of customers.  This would mean sub $50, which would be a bit more difficult.

hope that is all constructive
Thanks - very.