Up and Running

Apart from one intentional shutdown 25 days ago, PollyPi has been controlling my house without any problems for over a month, so I officially declare it to be fully operational!

What's next?

The first thing to look at is an intermittent problem where the response time of the system seems excessively slow.  The philosophy of putting all the control logic in the server and none in the remote nodes means when I press a button on a control box, a message has to go up and down before the button lights up.  This delay was never noticeable on the PC based system but it is sometimes obvious now.  The fact that it only occurs sometimes suggests it's not just a plain performance issue.  My gut feeling, based on no evidence whatsoever, is that the multiple writes to log files are clogging things up.  I can quickly confirm or disprove this by producing a test build which doesn't log anything.  Stay tuned ...

Progress At Last

I was finally able to devote a little time to the project, and I'm pleased to report that just a few minutes on the internet came up with a small tweak which some users reported had resolved their lockup problems.

So I added the magic incantation 

dwc_otg.speed=1

into the configuration file /boot/cmdline.txt and things are looking a lot better.

The up time is over four days at the time of writing, which is a lot better than it achieved before and I am currently letting Polly-Pi take over live control periodically, when a crash wouldn't be too inconvenient.

Now it's a bit early to be getting excited, all the previous incarnations of Polly have a total score of one crash in almost twenty years of continuous running, so four days hardly proves anything, but we'll see . . .

Update 09 April:  Still going strong. . .

Serial Port Woes

The project to move the server to run on a Raspberry Pi has ground to a halt.  I need two serial ports, one for collecting CLI information from the phone line, and more importantly, one for PICNET.  Using USB-Serial adaptors I have been experiencing lockups and crashes, there would appear to be a problem somewhere in the USB driver stack.  I have tried various models of USB adaptor but so far they all seem to experience the same problem.

Unfortunately I have been busy with other activities so I haven't been able to investigate in greater depth yet.  So far I've determined that my server software continues to run after these lockups, recording in its log file that the PICNET network has gone down and also recording loss of communications to the display client and all the ETHPIC nodes.  (The ethernet port on the Pi hangs off the USB system.)   So it would appear that the USB system has locked up in some way.

As a quick test I set up a level shifter for the on-board serial port and used that for PICNET and it worked fine.  I added a USB Serial adaptor, but didn't tell the Polly server to use it, and still there were no problems and a test run of ten days was completed.  Finally, without rebooting or changing anything else I configured the CLI function in the server software to use the USB adaptor, and within a few hours everything was locked up again, despite the fact that there had been no phone calls so presumably no serial data to transfer.  This same configuration - "real" serial port for PICNET and USB adaptor for CLI has been used on the previous server PC for years without difficulties.

So I think it's time to trawl the internet to see if anyone else is having problems.

Welcome To Polly-Pi

Welcome to Polly-Pi, where I intend to document the home automation system running in my house.  The site is still very much under construction, so please bear with me.