Project - EasyNet eXtra

LOGnode$ = "Log" 'Target name of error-logging node if available (useful if SD storage becomes available) ledpin = 13: pin.mode ledpin, output: ledoff = 1: pin(ledpin) = ledoff relay1pin = 12: pin.mode relay1pin, output: pin(relay1pin) = 0 'using active high qpio12 for relay1 buttonpin = 0: pin.mode buttonpin, input, pullup 'using active low gpio0 button interrupt buttonpin, pressed start=0: stop=0 'used by button-pressed subroutine to differentiate between short and long presses timer1 5000, Retry 'periodic timer to keep resending unACKed msgs until they expire udp.begin(udpport) onudp udpRX wlog "OK" wait udpRX: RXmsg$ = udp.read$ if ucase$(word$(RXmsg$,1)) = "ACK" then gosub ACK 'echoed reply from successfully received message, original msg can be removed from queue else target$ = ucase$(word$(RXmsg$,1)) 'Target may be NodeName or GroupName or "ALL" or localIP address if (target$=localIP$) OR (target$=ucase$(nodename$)) OR (instr(ucase$(groupname$),target$)>0) OR (target$="ALL") then instruction$ = trim$(ucase$(word$(RXmsg$,2))) 'Instruction is second word of message data$ = "": getdata data$,RXmsg$," ",2 'extract any data that follows the instruction if word.find(ucase$(instructionslist$),instruction$) > 0 then if (ucase$(instruction$) <> "ACK") and (instr(ucase$(data$),"ID=") > 0) then udp.reply "ACK " + RXmsg$ 'ACKnowledge the incoming msg endif gosub instruction$ 'branch to action the corresponding instruction subroutine else udp.reply RXmsg$ + " INSTRUCTION NOT RECOGNISED" endif 'word.find endif '(target$=localIP$) endif 'ACK return ACK: msg$ = "": getdata msg$, RXmsg$, " ", 1 wlog "Ack recvd for " + msg$ pos = word.find(retryq$,msg$,qdelimiter$) if pos > 0 then retryq$ = word.delete$(retryq$,pos,qdelimiter$) return RETRY: if word.count(retryq$, qdelimiter$) > 0 then if retryq$ <> "" then wlog "queue=" + retryq$ msg$ = word$(retryq$,1,qdelimiter$) 'grab first unACKed msg in the queue retryq$ = word.delete$(retryq$,1,qdelimiter$) 'chop msg off front of queue expire$ = "" WordParse expire$, msg$, "ID=", " " 'parse out ID= expire time if msg$ <> "" then 'compare expire time to current unix time if dateunix(date$) + timeunix(time$) > val(expire$) then 'Send LOGnode$ + "LOG ERROR: Node " + Nodename$ + " FAILED SEND - " + msg$ + " not ACKnowledged" else retryq$ = retryq$ + msg$ + qdelimiter$ udp.write netip$ + "255", udpport, msg$ wlog "retry " + msg$ endif endif endif return sub sendq(sendmsg$) sendmsg$ = sendmsg$ + " ID=" + str$(dateunix(date$) + timeunix(time$) + time2live, "%10d", 1) 'wlog sendmsg$ retryq$ = retryq$ + sendmsg$ + qdelimiter$ 'wlog retryq$ udp.write netip$ + "255", udpport, sendmsg$end sub sub send(sendmsg$) wlog sendmsg$ udp.write netip$ + "255", udpport, sendmsg$end sub sub GetData(ret$, v$, sep$, pos) 'extracts everything from the msg after the Instruction and puts into data$ (thanks cicciocb) local i, p, q p = 1 for i = 1 to pos p = instr(p + 1, v$, sep$) if p > 0 then p = p + len(sep$) next i if p = 0 then ret$ = "" else q = instr(p+1, v$, sep$) if q = 0 then q = 999 ret$ = mid$(v$, p) end if end sub sub WordParse(ret$, full$, search$, sep$) 'extracts value from option=value (thanks cicciocb) local p, b$ p = instr(full$, search$) if p <> 0 then b$ = mid$(full$, p + len(search$)) ret$ = word$(b$, 1, sep$) else ret$ = "" end if end subSTAGGER: 'used to provide staggered pause dependent on node IP to prevent multiple nodes responding together pause val(right$(nodeIP$,1)) * 10 return REPORT: gosub stagger msg$ = " Node name=" + Nodename$ udp.reply msg$ msg$ = " Groupnames=" + groupname$ udp.reply msg$ msg$ = " IP address=" + localIP$ udp.reply msg$ msg$ = " Commands=" + instructionslist$ udp.reply msg$ msg$ = " Date=" + date$ + " Time=" + time$ udp.reply msg$ msg$ = " RAMfree=" + str$(ramfree) + " Flashfree=" + str$(flashfree) udp.reply msg$ return REPLY: udp.reply "Response from " + Nodename$ return RESTART: reboot 'causes a device restart return RENAME: if data$ <> "" then nodename$ = data$ return LOAD: 'Load settings from file, by default is only looking for nodename, add your own saved parameters to look for if wished a$ = "" if FILE.EXISTS(filename$) > 0 then a$ = FILE.READ$(filename$) if WORD.GETPARAM$(a$,"nodename$") <> "" then nodename$ = WORD.GETPARAM$(a$,"nodename$") return SAVE: 'Save settings to file, by default will only save nodename, add your own parameters to save if wished a$ = "" if FILE.EXISTS(filename$) > 0 then a$ = FILE.READ$(filename$) WORD.SETPARAM a$, "nodename$", nodename$ FILE.SAVE filename$, a$ returnTIMEGET: udp.reply date$ + " " + time$ return TIMESET: msg$ = "settime " + data$ command msg$ return Relay1ON: pin(relay1pin) = 1 return Relay1OFF: pin(relay1pin) = 0 return Relay1Toggle: if pin(relay1pin) = 1 then pin(relay1pin) = 0 else pin(relay1pin) = 1 returnBLINKNODE: ledstate = pin(ledpin) blinkon = 200 blinkoff = 400 blinkpause = 1000 blinkgap = 1400 pin(ledpin) = ledoff pause blinkpause for pos = 1 to len(nodeIP$) digitchr$ = mid$(nodeIP$,pos,1) if digitchr$ = "0" then digit = 10 else digit = val(digitchr$) for count = 1 to digit if ledoff = 0 then pin(ledpin) = 1 else pin(ledpin) = 0 pause blinkon if ledoff = 0 then pin(ledpin) = 0 else pin(ledpin) = 1 pause blinkoff next count pause blinkpause next pos pause blinkgap pin(ledpin) = ledstate return BLINKIP: ledstate = pin(ledpin) blinkon = 200 blinkoff = 400 blinkpause = 1000 blinkgap = 1400 pin(ledpin) = ledoff pause blinkpause for pos = 1 to len(localIP$) digitchr$ = mid$(localIP$,pos,1) if digitchr$ = "." then pause blinkgap else if digitchr$ = "0" then digit = 10 else digit = val(digitchr$) for count = 1 to digit if ledoff = 0 then pin(ledpin) = 1 else pin(ledpin) = 0 pause blinkon if ledoff = 0 then pin(ledpin) = 0 else pin(ledpin) = 1 pause blinkoff next count pause blinkpause end if next pos pause blinkgap pin(ledpin) = ledstate return BLINKS: blinksnum = 5 ledstate = pin(ledpin) pin(ledpin) = ledoff pause 200 if data$ <> "" then blinksnum = val(data$) for count = 1 to blinksnum if ledoff = 1 then pin(ledpin) = 0 else pin(ledpin) = 1 pause 800 pin(ledpin) = ledoff pause 200 next count pause 2000 pin(ledpin) = ledstate 'Restore LED state to its previous state return PRESSED: if pin(buttonpin) = 0 then start = millis else stop = millis if stop > start then if stop - start < 2000 then send "All blinks 12" 'short press else send "ALL BlinkIP" 'long press endif endif return END '-------------------- End ---------------------

UDP has no native handshaking or arbitration, so EasyNet achieves its own handshaking by sending via its RetryQ$ which will keep resending the message until the Target returns ACKnowledgement of receipt, or the msg times out.

It also offers optional rudimentary arbitration whereby multiple node responses to a message can each be slightly delayed according to their unique node address.

I considered a minimal pause between each node transmitting should be 10 milliseconds, but I was mindful that using the full node address for the pause could give delays potentially ranging from (001 x 10 =) 10 millisecs, to (254 x 10 =) 2.54 seconds (which would be acceptable if there were a lot of nodes responding), but for the moment I am only using a handful of nodes with node address in 70;s and 80's. So I opted to use just the last digit of the node address,which proved effective for my needs -, but you may wish to create a version tailored more to your own needs.

.

In the script above, it is assumed the new "REPORT" instruction may be issued to "ALL" nodes (ie: All Report), so to prevent all nodes from swamping each other by trying to respond at the same time, the first line of the REPORT subroutine is a GoSub STAGGER, which pauses for a delay which is defined by the last digit of the node address x 10 millisecs.

STAGGER: 'used to provide staggered pause dependent on node IP to prevent multiple nodes responding together

pause val(right$(nodeIP$,1)) * 10

return

There are undoubtedly better ways to acheive a staggered response, but I've been trying to avoid too much unnecessary clutter to leave room for other features which I wish to add.

Planned Enhancements

A WatchDog mechanism could prevent nodes disappearing offline without being noticed, but it is still only at the planning stage at the moment.

The plan so far is to add a new EasyNet instructions called WatchDog, whereby any node can send a WatchDog instruction to any other node to make it assume the responsibility of being a watchdog monitor for it.

A prospective client would send a designated watchdog target node a WatchDog instruction which includes its required wakeup duration plus its obituary$ to be actioned if it fails to respond. The obituary can be anything that can be recognised and actioned locally by the watchdog node, or anything that can be recognised and actioned by any remote EasyNet node(s), and will include ability for sending multiple msgs to multiple targets.

When a WatchDog instruction is received, it adds the sender to a local watchlist$ of watchdog clients along with its specified obituary msg and next wakeup time, calculated by adding the clients specified wakeup duration to the local unix time to give a local time for when the wakeup call is due.

The first implementation will probably use a variation of the RetryQ to monitor the clients, periodically checking the first entry in its watchlist$.

If the wakeup time is not yet due, the entry will just be snipped off from the front and added to the end of the list - quick and simple.

But if the first entry wakeup time is due, the entry will be deleted from watchlist$ and a wakeup call sent to the client.

The wakeup call will probably just be another new EasyNet instruction called WakeUp which will issue another WatchDog instruction requesting a new wakeup call, thus allowing clients to keep requesting new wakeup calls indefinitely for as long as they are able to keep responding.

The WakeUp subroutine might be something like...

WakeUp:

Watchdog$ = "Watchdog" 'Nodename or partial Groupname of a designated watchdog node (there could be more than one watchdog).

WakeupDuration = 10 * 60 '10 minute keepalive duration

Obituary$ = "SPEAK Warning, Node " + Nodename$ + "is no longer in contact" 'RIP message to be sent if communications lost

SendQ WatchDog$ + " WatchDog " + Nodename$ + " " + WakeupDuration + " " [+ "RIP=" + Obituary$] 'clients watchdog request

return

Where:

Watchdog$ is the Nodename or partial Groupname of the required watchdog node (there could be more than one watchdog).

WatchDog is the EasyNet watchdog instruction recognised in the instructionslist$ of a watchdog 'server'.

Nodename$ is the client nodename requesting to be added to the watchdogs watchlist$.

WakeupDuration is the delay before the client requires a wakeup call (individual clients can specify their own individual wakeup durations).

Obituary$ is the final message(s) for the lost nodes which is used to flag up its loss when it cannot be contacted.

When the target node replies, the new WatchDog details will overwrite and update the previous expired watchdog details, and the ACKnowledged sent message will be removed from the RetryQ.

But if the target node fails to respond, the sent message will eventually expire from the RetryQ, causing the watchdog node to action the failed nodes RIP=Obituary$ message(s).

If a watchdog client node wantes to have its watchdog monitoring ended without triggering its obituary, it could send a WatchDog message without any wakeupduration, which could act as a request to be removed from that watchdogs list.

In summary, any node can request any other node to be its WatchDog monitor to send it WakeUp calls at specified durations.

If a the client fails to respond before the sent message expires, then the WatchDog will notify of the clients loss by sending the clients 'Obituary'.

This could be implemented on just 2 nodes, with each acting as the watchdog for the other, perhaps blinking an LED and bleeping a buzzer in the event of lost communications with the other node.

In my case I would want my Voice Announcer node to keep periodically speaking the nodes failure until I acknowledged the problem.

I would probably also wish to give visual indication of the failure on my sensor Alerts Monitor.

Eventually also to log the failures to SD after I have created a planned ErrorLog node.

The important point is that any node could request any other node to be its watchdog and issue it wakeup calls for specified durations then to take the specified obituary$ action(s) in event of failure.

Therefore a watchdog node could designate another node to act as watchdog for it.

Thus it would be possible to monitor all nodes, and for mission-critical nodes to have backup devices waiting to take over important duties.

Even with the new WatchDog and WakeUp instructions available for all, they would be optional, and would be specifically invoked if required.

Only when a client node issues a WatchDog instruction will it be put on a watchlist$ to receive wakeup calls.

And only when a designated node receives a WatchDog instruction will it have any clients in its watchlist$ to send wakeup calls to.

Well, that's the plan, anyway... but I am currently trapped into disaster recovery, so don't know when I will get opportunity to implement and test it.

When a sender deletes an UN-acked msg from its retryQ, it could also broadcast an appropriate error msg to a LOG node if one exists (perhaps using the optional RIP=obituary$ flag).

It could do it anyway... to cater for a planned future LOG node - because until something listens, the error messages merely fall on deaf ears.

Annex does not have SD capability yet, so a current alternative might be to use an 'external' serial data-logger resource, such as arduino.

This could be quite easy using a Wemos D1 Mini 'dual' or 'triple' base shield, allowing an Annex EasyNet node to sit alongside an arduino firmware wemos with a plug-in SD or data-logger shield.