Strange reset during the day

[TomW]

If the Local App quit communicating with the Classic, make sure the Classic itself is running
and charging batteries, which is its main function, but the communications is very important
too.

boB

boB;

I don't want to cloud the waters but..

I seldom use the Local App but mine both seem to work just fine as controllers and seem to display correct KWH totals before the reboot required to recover communications . I will take a closer look when it happens again.

When I did use the L.A. the communication fault required a reboot to reconnect and of course the data resets.  I never noticed it reset while the L.A. was able to communicate with the Classic?

Do people see the data reset with no L.A. connection issues?  My logs do not seem to show data loss or resets other than from the recovery power cycle?  If that makes sense?

Curiouser and curiouser.

Tom

[zoneblue]

I dont use my midnite, nor the local app, and currently running 1370. Just use newmodbus and related code. Here im pretty sure ive never seen a full spontaneous reset, and can verify that from the logs. And newmodbus was running perfectly on an intel 686 platform  for some months. However trying to use it on an embedded arm board, results in the above pattern of closed ethernet port about once a day.

Most times newmodbus will timeout waiting for a response but other times the classic will refuse the connections. It will do this sporadically on and off for few hours, then finally closing completely.  At that point the local app will not be able to connect if you try, either. The classic appears to continue to operate as normal, just if you want to talk to it, youll have to reboot it.

Its actually very very reminiscant of how the local app used to work with 1070.

Now im mindful that i have 10m of 5e betw the classic and switch. And that the insides of a charge controller is probably one of the noisiest environments known to man, so i will try adding another switch closer to the classic. And removing the cable tie restraint i have from the 5e to one of the power cables.

Why it works on one computer but not on the other is so far still a mystery to me. But i will say this, that battling the ethernet port on the classic is starting to wear me down. I NEED reliable data out of the thing. The combination of classic and arm board is to me the optimum solution from a power consumption point of view. If i need to add another switch, sigh, i can live with it. Its still less than the 20W from the intel fileserver, which runs, along with the wifi, during daylight hours / on demand only.

To rule out newmodbus, youll recall i have previously tried using phpmodbus and the python modbus library. All exhibit the same pattern.

[RossW]

use newmodbus and related code. Here im pretty sure ive never seen a full spontaneous reset, and can verify that from the logs. And newmodbus was running perfectly on an intel 686 platform  for some months. However trying to use it on an embedded arm board, results in the above pattern of closed ethernet port about once a day.

To rule out newmodbus, youll recall i have previously tried using phpmodbus and the python modbus library. All exhibit the same pattern.

Interesting.

Tom is having similar issues, and is running a raspberry pi.
He has compiled *exactly* the same source code that I have, yet I've never had loss of communications to my classic

I'm running newmodbus on a FreeBSD box (local home server), CPU is:
   CPU: VIA Esther processor 1200MHz (1200.01-MHz 686-class CPU)
      Origin = "CentaurHauls"  Id = 0x6a9  Family = 6  Model = a  Stepping = 9

The classic and several other ethernet devices are out in the battery room, connected (by choice - "air gap" isolation) a WRT54GL reflashed with DD-WRT. This gives me a local 5-port switch and wireless bridge in one convenient box, so all the physical ethernet cables there are standard 1.8 or 2m cables - not screened or specal, just cheap patch leads.

Perhaps newmodbus needs compiled with different compiler optimisations? Any suggestions?

[TomW]

Why it works on one computer but not on the other is so far still a mystery to me. But i will say this, that battling the ethernet port on the classic is starting to wear me down. I NEED reliable data out of the thing. The combination of classic and arm board is to me the optimum solution from a power consumption point of view. If i need to add another switch, sigh, i can live with it. Its still less than the 20W from the intel fileserver, which runs, along with the wifi, during daylight hours / on demand only.

To rule out newmodbus, youll recall i have previously tried using phpmodbus and the python modbus library. All exhibit the same pattern.

Zone;

Well, that kind of makes me wonder.

Guess I will put another X86 computer online, disable the PI logging and see if I can get steady data with that a few days.

Color me confused. I guess "standards" are not. Anyway, interesting angle Iwas touching on in my quest for what is going on.

And, I agree, battling this issue has gotten on my last nerve.

More info if / when I find it.

I see I sat on this a bit too long and Ross chimed in but here it is anyway

Tom

[zoneblue]

Putting this all in context, i am close to a "proof of concept" release  of come code for "the black box" project. Ive basically taken the quick and dirty php i initially coded, that was necessarily tailored  to the handful of registers that interested me, and refactored it to be web ui configurable. So you can choose which datapoints to show on what part of the screen, what size and style, and which graphs to show, size, series datapoint(s), color,  etc.  Another week or 2 til theres something to see.

Ross. the x86 box that works is an intel atom dual core.  When i said that the coms issues occured with phpmodbus as well, that was with 1070 and the atom. I havent tried python modbus on the cubie, but sounds like i ought to. I also need to try 1549 , but i gather there is another coming, so will wait for that. Hours in the day.

[RickW]

Just put a Classic 150 in service a week ago.  Like others, it resets about noon.  Not sure about other times as I am prepping for winter and not around the equipment during the day. 

I did the basic programing and let it go.  There is a date somewhere in the menu of 12-01-12 but just going off memory.  Was just getting ready to find the latest firmware, thinking that must be it but not sure now?  Not running any networking or apps, just the 150 hooked up doing its thing.  Guess I will watch this thread?

Rick

[Halfcrazy]

Rick
If it is resetting around noon my money is the clock is set incorrectly. The Classic uses military time or 24hr clock. This has gotten me many times. I would check and see if the Clock is set wrong.

Ryan

[boB]

The Classic won't do a hard reset once a day (midnight) unless the A-RST or Auto Restart is turned on
in Tweaks.  However, at midnight, every day, (or what the Classic thinks is midnight 23:59) it will
zero out the daily kW-Hours, add it into the Life-Time kW-Hours and start a new bulk cycle in
preparation for the next day.

Also, if for some reason the "Time Synch" gets turned OFF in Tweaks, the MNGP will stop
setting the time in the Classic itself.   This Tweak is usually on unless turned off though
but might be worth a check.

boB

[RickW]

OK Ryan, I feel like a real dummy.  I did indeed have the clock set wrong.  My bad. 

Glad you mentioned the Time Synch, boB.  I'm trying to understand the Classic.  After coming from 8 years with the MX60 I'm feeling my way around the Classic.  Gotta learn these things because I have several more of the 150's to install!

Thanks for your quick response gent's.  By the way, the firmware update instructions are right-on, appreciate the detailed instructions. 

Rick

[boB]

OK Ryan, I feel like a real dummy.  I did indeed have the clock set wrong.  My bad. 

Glad you mentioned the Time Synch, boB.  I'm trying to understand the Classic.  After coming from 8 years with the MX60 I'm feeling my way around the Classic.  Gotta learn these things because I have several more of the 150's to install!

Thanks for your quick response gent's.  By the way, the firmware update instructions are right-on, appreciate the detailed instructions. 

Rick

Excellent, Rick !  Glad you got it going OK.

The old MX60 is a good controller, too.

boB

[Resthome]

Just as another data point I have been running 1370 firmware on the classic for some time along with Local App V0.3.27 with Web Access Disabled and have not come across any in day resets. I do an Auto Reset at midnight for what it is worth. Classic is networked through a RadioLabs USB Wifi Router Repeater - WiJacker with a CaptiFi USB WiFi antenna from RadioLabs.

[zoneblue]

Just as an interesting aside,  on a lite,auto reset doesnt really work, or works rather interestingly.

Im using it at the moment to resolve a closed ethernet port issue, but becasue the lite doesnt have a RTC, its tricky to get it to restart at midnite. Nominally it appears to reboot at 00:00:00 but im not sure what happens but it seems to end up rebooting both morning and evening all sorts of odd times. I havent figured the pattern out yet, but am monitoring it.

[boB]

We are supposed to have a local app soon that allows you to set the time/date on
the Classic.

Guess it didn't quite make the latest one though.  Maybe next one ?

Thanks for the info !  Still working on this.

boB

[aroxburgh]

Bob:

I updated soon after 1401 was made available for the Classic. Since then I have had no resets! (not including the A-RST = on midnight resets)! Great job!!!!
Other than moving (temporarily to a worse solar aspect ~0.3 kWh per day, and now to a much better one than previously, ~3 kWh per day ), there have been no significant changes to my hardware setup/situation.

I have not used the local app much during this time, preferring not to run my PC around the clock to log the data (which also goes against the grain since the data is already logged internally in the Classic, so any good local app should be able to access that, which the current one does not).

Also in the time I've been using 1401, I have not saturated my Internet connection with other traffic, a situation previously reported to you as causing resets.

FYI, my generation for the past week (1st week of Sept 2013) is as follows:

Date         Energy           Peak Power
2 Mon       2.5 kWh           1095 W   (= 94% of array rating).
3 Tue        3.0 kWh            923 W
4 Wed      1.9 kWh          1071 W
5 Thu       1.0 kWh            201 W
6 Fri         0.7 kWh            262 W
7 Sat        2.9 kWh            794 W

BTW, is the absence of resets in 1401 an accident, or was a specific bug identified and fixed?
Right now I am very happy with performance of my system. The only thing on my current wish list is remote access of the Classic's internally logged data from any browser. ;-)

Al
aj4rf

[zoneblue]

The only thing on my current wish list is remote access of the Classic's internally logged data from any browser. ;-)

Well theres mymidnite. Soon there will be the blackbox project as well. Its a tiny computer that can talk to the classic, and do far more extensive logging than the classic alone can do, owing to its finite memory. Draws a little over 1W, and can store a whole lifetime of realttime data. Watch this space. Another week or so maybe.