(FIXED) IMPORTANT: Issues after software upgrade and river inspection 19/04/2018

[hamishmb]

Hi all,

When I came in on Thursday we had a number of issues, the main ones being:

• The sump had lost almost all water.
• After upgrading the pis to the latest version of Raspbian Stretch (9.4 I think), the software didn't start automatically.
• The butts pump kept turning on and off every few seconds.

1) I'm not sure why we lost water - perhaps someone ccan have a look at the logs and figure this out?

2)

Both pis exhibit this behaviour.

As for why this happened, I couldn't say. The software works when I start it manually, but either boot up isn't completing, auto-login isn't working, or .bashrc isn't being executed on login.

I tried removing Terry's tty check, but it didn't fix this issue.

I also ran "systemctl status", which told me the state was degraded, and a bluetooth-related service had failed to start. I doubt this would block login but it's possible.

Terry, how did you configure auto-login? I have a hunch as to how to fix this, but we'll need to connect at least one of the pis to your monitor to see exactly what is going on I think. I think maybe we should be starting the software using /etc/rc.local instead of /home/pi/.bashrc, or perhaps some configuration was lost during the upgrade. I kept backups of the original software, but I am reluctant to downgrade because that' just delaying the issue rather than fixing it.

Terry, could you have a look at this for me because I'm not sure how you configured auto login and other things in the first instance, so I don't wwant to mess about until we know what's gone wrong.

3)

I couldn't determine the cause of this issue. Terry, are you sure we plugged the sumps into the right plugs in the SSR?

Also, I have gone and patched the disabled side of the river, but I think we'll need to drain the section by the sump very soon to fix the main problem - we know there were leaks there too.

Because of these issues, I have disabled the river control system, but the river is still running. Sorry about this!

Hamish

[TerryJC]

• After upgrading the pis to the latest version of Raspbian Stretch (9.4 I think), the software didn't start automatically.

I had problems similar to this when I upgraded the Bells System to Stretch and I wasn't using the .bashrc technique. I can't remember exactly what I did, but it was much easier with the Bells because most of the documentation was complete.

You can see the latest versions of the User Manual and Installation Spec for the Audio Guide and Town Quiz at https://wmtprojectsforum.altervista.org ... &t=20#p135. As pointed out by Penri at the time, these documents cover the Bells setup too.

Terry, how did you configure auto-login? I have a hunch as to how to fix this, but we'll need to connect at least one of the pis to your monitor to see exactly what is going on I think. I think maybe we should be starting the software using /etc/rc.local instead of /home/pi/.bashrc, or perhaps some configuration was lost during the upgrade. I kept backups of the original software, but I am reluctant to downgrade because that' just delaying the issue rather than fixing it.

I can't get in to connect the monitor today until around 4 pm at the earliest and trying to do it this weekend might be problematic if the site is busy. Can you make it at 4 pm today?

Auto login is disabled by default; you have to disable it to stop it happening, so I doubt that is the problem. Using /etc/rc.local to start the software is fine, and would remove the need to have the Pi login at boot-up, but it won't solve the problem if the Pi isn't setting itself up properly in the first place.

It may be easier to downgrade and then upgrade the your software to check that it is not causing an issue before trying to upgrade Stretch. Do we really need to upgrade Stretch at this time? Unless there is some major advantage, it may be quicker and easier to stick with the version we had on Tuesday and just add your new program.

Terry, could you have a look at this for me because I'm not sure how you configured auto login and other things in the first instance, so I don't want to mess about until we know what's gone wrong.

As you say, the only way to see what is going on at boot-up is to have a monitor connected. I can come in later as mentioned above.

3) I couldn't determine the cause of this issue. Terry, are you sure we plugged the sumps into the right plugs in the SSR?

Yes. We even tested it and observed the pumps starting and stopping.

[hamishmb]

The software is fine - it works if I run it manually through ssh.

I'd rather have it up to date (ish) because that's what I'm testing against at home - my pi is continually self-updating. Also, if I update it using my laptop (chroot) fairly often then it will take less long each time XD.

I can come in at 4 PM, shall we do that then? I imagine it will come clear when we connect the monitor. Note we were already on stretch, just an old version, so this is a bit weird.

Hamish

[TerryJC]

Hamish,

I didn't think that your software was the problem; but you ended up changing several things at the same time; always a risk.

I'll see you at 4 pm.

[TerryJC]

OK.

Just got back from WMT. The main problem was that the upgrade to Stretch had disabled auto-login.

I think this might be because we have SSH enabled (they are paranoid about changing the password if you enable SSH).

We also found that the Main Circulation pump was on when it should have been off. We set it to On for the moment and will debug when we've checked out all the software and hardware.

[TerryJC]

Hamish,

I've just checked the circuit diagram, Interfaces Board Layout and Wiring Schedule and as far as I am concerned, the active GPIO pin for SSR2 is pin 18. The Level Converters are non-inverting and so, barring errors when I did the wiring mods on Tuesday, setting GPIO 18 High should turn the Main Circulation pump on and setting it Low should turn it off.

If your software is doing the right thing, then we'll have to checkout the operation with a multimeter.

[hamishmb]

Terry,

That's in accordance with the software, and I can confirm that for Motor objects (SSRs) the active state is high, and that I've configured the main pump in the same way as the butts pump. Maybe we do need to take a multimeter to it.

Just so Penri knows what's happened:

We have fixed the issue I mentioned yesterday where the software wasn't starting - the auto-login config was lost.

However, the SSR / some of the circuitry / software appears to have a fault because the circulation pump doesn't turn off when it should. We may also have an issue with the butts pump - aforementioned "pulsing" on and off every few seconds. We can investigate this on Tuesday I suppose. We have left the system in its previous state, minus control of the circulation pump.

On another note, the real time clock is wrong, but it may just need to be reset.

Terry - I will look at what that auto login setting changes, so when I update the software again, I can manually set it from the commandline

Hamish

[TerryJC]

On another note, the real time clock is wrong, but it may just need to be reset.

That will need looking into. It should have kept going independent of the update.

[hamishmb]

Hmm okay. I'm sure we'll live without it for now XD.

Also - thanks for coming in and helping me fix this Terry

I think I see how the auto login feature works. I shall experiment on my pi, so I know how to re-do it after updating the pis through my laptop.

Hamish