Crashed Instance and dissapeared production flows due to "version mismatch"

Support
21

Thanks for the feedback.

This is not the FlowForge version unfortunately.

Please see the admin/overview

image
image699×664 25.1 KB

22

Overview FlowForge version V1.6.0

23

The merge feature is part of Node-RED. And following discussion with @rob I am told it is very reliable.

In fact, as a maintainer of Node-RED myself, I have not seen this nor have I seen any reports of an issue.

If you can re-create the problem in a repeatable mannor - or details the exact steps that led to the crash - I will look definately into this.

In the mean time, I will discuss the merit of some kind of auto-snapshot feature with my collegues.

1 Like
24

I understand. So the issue is still unresolved.

Thanks again for your kindly help.

We will need to go ahead to perform the changes again and check for ways to avoid this in a near future from our end, we would love to recreate the issue, the information we have is already on the thread.

25

@jvalencia - I hope you had a less eventful day today :slight_smile:

I note you are on V1.6.0.

With V1.8.0, we added improved pipeline features that may help in these situations.

It is procedural (not automated) but could be used to defend against future issues.

Please check it out: https://www.youtube.com/watch?v=Pbql22f3vqY

Also, please remember, if you can create a repeatable set of steps to crash node-red as you did, please let me know.

26

@Steve - Thank you for reaching me again! We hope that too :smiley:

You are correct, we were using V1.6.0 however after this issue we decided to upgrade and now we are using V1.8.2 so that the pipelines may help in these situations. It doesnt matter if its only procedural by now, as it will help to defend again future issues.

We will look forward to watch the video and see how to integrate it in our flow.

Let me get more information with the team so that we provide you with the set of steps to crash node red as we did :smiling_face_with_tear:

Also, @zackwasli propossed a meeting to discuss an upgrade to a paid tier version, and we setled to the next monday.

27

Here are the steps we think may caused the crash.

  1. We were following the deployment in 2 different accounts and PCs, “A” is admin’s where the hard work is happening, “B” is a basic user permission following the deployment, sending triggers, creating debbug nodes, etc.

  2. “B” created a debug node and disconnected a node, basic operations to debugging

  3. Seems like “A” created a deploy before receive the merging notification and sent it while the one in “B” was still happening.

  4. We think, after that, “A” did received merge notification, accepted (Its our supposition as we dont recall the exact situation) … then the system crashed and the flows dissapeared.

We concluded that the system received 2 deployments at the same time, due to the industrial environment the latency of the network is sometimes very high, and we suppose NodeRED was unable to catch the differences before creating the merge notification.

Please let us know if that is enough for you. As we stated, seems the NodeRED logs were also deleted as we didnt find logs for that moment, logs were for 20, 30 minutes after the crash.

Kind regards.
Jose Valencia.
Matec Software Development Engineer.