Errors and `node-id` values

Ok, I have been down this road before - sorry.
I have a SHOCKING memory to what was resolved from my question.

What I think:
All nodes have a unique id so when they have an error it is given and you/I can search for that node.
I say example, but they are from real errors that happened on different machines.
The format is usually like this: (example)
90c309f3.55a2

I asked a while back because I was getting errors with no . in the node's ID field.
They look like this: (example)
e4cf8781c0c037bf

Yet, today I see that this IS valid.

Could someone who is up to speed with all this mind helping me understand what is what?
Thanks in advance.

The format changed, with node red 3.0 I think, but not certain about that.

1 Like

Appreciated.

I have to get my devices up to date, but some are so old they are still on STRETCH!
(Yeah, I know.)

But the main ones are on Buster, so that is good.

If I am on 2.1.4 and 2.2.2:
Will that be ok to update to 3.x (running Buster on the machine)?

Just to note that updating Node-RED will not change the ids of any existing nodes - only newly added ones.

1 Like

Thanks.

I am just maybe accepting that I should update the main machines I use nearly DAILY to 3.x rather than having a mish-mash of versions.

I just don't want to break that which is working after all the fun I've just had with stupid mistakes I've made in other areas.

It has cost me weeks trying to get things working, and I have only just got things nearly back to a happy state.

If a machine is doing its job and I am not doing further development on it, and it is not exposed to the internet, then my policy is to leave it alone.
If I have to do any significant new work on it, however, then I upgrade it first, make sure it still works, and then start the new development.

2 Likes

Hi,

And take a backup first, in case of emergencies... right?

Cheers!

Paul

2 Likes

I keep image backups of all my PIs, so if I mess up or lose a card (or pi) I can be up and running again very quickly.

Other than taking the SD card out of the machine and backing it up on another machine to another card.....

Can it be done in situ? (in place)

That is the way I do it (using clonezilla, which gives relatively small images as it only images the used sections of the card, not all the empty space.).

I don't know.

1 Like

Ok, thanks.

But that means the device (RasPi) has to be taken off line while backing up.

Yes. But you need another device of at least the same size. So on a Pi, a USB stick would do the trick. You can use the age-old dd utility to do an exact backup of any storage device on Linux.

Well, actually that's not true! If you have a large SD-Card you could partition it (not a lot of actual help of course!). Or you can dd to a file and move that off-device.

But to do dd the partition has to be unmounted first - doesn't it?

(Not having a go at you.... Just that is something else I only just remembered now.)

So basically I'm stuck between a rock and hard place to make active backups.

Clonezilla looks like a good option if it makes small images.

I am pretty sure it does need to be unmounted, as if a file is being written to as dd is reading that files area the file (or folder) could be corrupted in the image.

1 Like

Seems I am really stuck with this then.

As said: I may just have to go with clonezilla as suggested.
Just taking the main machines offline to to backups is not ..... good.
But if it has to be done, then so be it.

Neither is finding that the SD card has gone AWOL and you have no easy way to build a new one.

1 Like

Agreed.

I do backup my flows to a USB stick now and then (only when things change) so that does reduce a lot of heartache.

I need to get smarter so NR monitors / detects new backups (watch node?) and copies them to the USB stick.

Ok, quick explanation of why I'm doing it that way:
The USB is formatted NTFS. (Good and bad)
The SD card supports sharing and I share the public directory so I can copy any backups to a REMOTE machine. I still haven't got that working as good as I would like.
So as a second option I also copy them to the USB stick.
Then if the SD card goes fft! I have a backup on the USB stick.

This is a whole other can of works - active backups - so I shall not carry on about that here.
When I get time/motivation I will have to look at getting something working that keeps active backups.
But that in itself may need to be backed up now and then..... argh! :wink:

Come on folk, the answers are only a search away. How to Clone Your Linux Hard Drive: 4 Methods (makeuseof.com).

dd on a mounted drive has some potential risks but it has been the go-to tool for decades. But as shown, there are loads of other tools too. The Linux filing systems are very robust and handle this kind of thing very well.

Depending on how complex your OS config is. An easy answer is to keep a pre-configured spare to hand and only back up the node-red (and db maybe) essentials. I've shared some scripts for node-red backups before and you can simply extend to zip up the latest and SCP/FTPS it to another device.

Recovery is as simple as copying over the latest node-red backup, swap the card and go.

Exactly, that was my point.
I use git to backup but the effect is the same.