# Issues using $parent. from a subflow

**URL:** <https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231>\
**Category:** General\
**Created:** [17 March 2020 14:31 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231 "2020-03-17T14:31:42Z")\
**Posts on this page:** 16\
**Page:** 1

<div class="post-metadata">

**Author:** ![nileio](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/nileio/32/9444_2.png) [@nileio](https://discourse.nodered.org/u/nileio)\
**Post date:** [17 March 2020 14:31 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/1 "2020-03-17T14:31:42Z")

</div>

Hi all,  
I am facing issues with using `$parent.var` from a subflow in order to access the parent flow context. Though my context variable exists, the subflow does not return any value. It does not even return `undefined`.

the problem is that it works in some cases and does not work on others.

Is there any trace information I can enable or so to find out what actually happens when `$parent` is being used in a subflow and why it may behave this way when there is definitely a context var?

the only solution guaranteed to work at this stage is if I create the subflow from scratch. If I import a subflow or create a subflow from exisiting nodes , well for that , actually sometimes it works (in very simple flows) and other times it does not ! I could not for certain reproduce it in all situations, therefore my question.

Thanks for help.

---

<div class="post-metadata">

**Author:** ![knolleary](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/knolleary/32/3_2.png) [@knolleary](https://discourse.nodered.org/u/knolleary)\
**Post date:** [17 March 2020 14:47 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/2 "2020-03-17T14:47:37Z")

</div>

Hi @nileio

there is no specific tracing in this area of code.

If you are able to narrow it down at all, or at very least give an example sequence of steps that does reliably reproduce it, that would be very helpful.

---

<div class="post-metadata">

**Author:** ![nileio](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/nileio/32/9444_2.png) [@nileio](https://discourse.nodered.org/u/nileio)\
**Post date:** [17 March 2020 15:03 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/3 "2020-03-17T15:03:53Z")

</div>

hi @knolleary  
I originally had a massive flow then I started breaking down into subflows and changed references to use `$parent` rather than accessing context directly. Everything worked fine on first few tests. I noticed after restart and took me a while to find out that my flow is failing because of the nodes using `$parent.`  
The way i can hardly reproduce is

1. Have a working flow then create a subflow from the nodes.

2. some nodes to create a flow context vars (might be `inject`-`change` to test but mine is produced by websockets)

3. test `$parent` from the subflow using a simple `Change`-then `Debug` nodes . It works fine ? then restart nodered, and test again.

I am sorry that I cannot for definite reproduce it but its happening in my environments and I ran out of any possible reason. Almost recreated my whole flow twice !  
The thing is if I create a subflow from scratch and copy-past few nodes at a time , then deploy, it seems to work. However, my subflow is large and it is not feasible for me to do that.  
The following is an exact replica example of what I have but the difference is in my case the subflow have far many nodes. This one here will just work fine.. but mine will not though they both try access to same context var `somecontext`

```auto
[{"id":"e8c7b15d.e5a8e","type":"subflow","name":"subflow","info":"","category":"","in":[{"x":450,"y":300,"wires":[{"id":"30f1cd29.67ea92"}]}],"out":[{"x":740,"y":300,"wires":[{"id":"30f1cd29.67ea92","port":0}]}],"env":[],"color":"#DEB887"},{"id":"30f1cd29.67ea92","type":"change","z":"e8c7b15d.e5a8e","name":"","rules":[{"t":"set","p":"payload","pt":"msg","to":"$parent.somecontext","tot":"flow"}],"action":"","property":"","from":"","to":"","reg":false,"x":610,"y":300,"wires":[[]]},{"id":"5ebfa62a.9466a8","type":"tab","label":"Test parentContext","disabled":false,"info":""},{"id":"de134d62.2a6ab","type":"inject","z":"5ebfa62a.9466a8","name":"","topic":"","payload":"","payloadType":"date","repeat":"","crontab":"","once":false,"onceDelay":0.1,"x":515,"y":305,"wires":[["70b61b.1c5fc9e4"]]},{"id":"70b61b.1c5fc9e4","type":"change","z":"5ebfa62a.9466a8","name":"","rules":[{"t":"set","p":"somecontext","pt":"flow","to":"test","tot":"str"}],"action":"","property":"","from":"","to":"","reg":false,"x":705,"y":305,"wires":[[]]},{"id":"3456050e.bbd32a","type":"subflow:e8c7b15d.e5a8e","z":"5ebfa62a.9466a8","name":"","env":[],"x":720,"y":370,"wires":[["77bdd3d8.323e5c"]]},{"id":"48785205.3094ec","type":"inject","z":"5ebfa62a.9466a8","name":"","topic":"","payload":"","payloadType":"date","repeat":"","crontab":"","once":false,"onceDelay":0.1,"x":570,"y":370,"wires":[["3456050e.bbd32a"]]},{"id":"77bdd3d8.323e5c","type":"debug","z":"5ebfa62a.9466a8","name":"","active":true,"tosidebar":true,"console":false,"tostatus":false,"complete":"false","x":880,"y":370,"wires":[]}]

```

In order to test this against my problem subflows, I tried with one subflow and removed all connections from the `input` node of the subflow to the rest of my nodes and just connected the `input` node with simple `Change` then `Debug` as per the example. To my surprise, when I test by using the `Inject` node on the root flow nothing happens ! I dont even get `undefined` in the sidebar. It is a bit weird behaviour but this is happening in 4 sub flows that I currently have..

---

<div class="post-metadata">

**Author:** ![bakman2](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/bakman2/32/6207_2.png) [@bakman2](https://discourse.nodered.org/u/bakman2)\
**Post date:** [17 March 2020 17:19 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/4 "2020-03-17T17:19:37Z")

</div>

within the subflow:

to retrieve the flow context of the parent, there are multiple ways:

in a function node  
`flow.get("$parent.somecontext")`

the change node, and how you set it up  
or with a change node, using the jsonata expression option with:

 ![Screenshot 2020-03-17 at 18.18.49](https://us1.discourse-cdn.com/flex026/uploads/nodered/original/3X/7/a/7a2e108b54b175750e0e5de7cccd36c363baa8ef.jpeg)

Make sure to initialize the flow variable upon restart else it will become undefined, you can do this by changing the inject node by enabling the checkbox "Inject once after x seconds, then" or use persistent storage for the variables in settings.

---

<div class="post-metadata">

**Author:** ![knolleary](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/knolleary/32/3_2.png) [@knolleary](https://discourse.nodered.org/u/knolleary)\
**Post date:** [17 March 2020 17:27 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/5 "2020-03-17T17:27:40Z")

</div>

Hi @bakman2

your post appears to be explaining how to use `$parent.` - does it actually address the issues that @nileio is hitting? Have you spotted something in his example flow that is wrong?

---

<div class="post-metadata">

**Author:** ![bakman2](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/bakman2/32/6207_2.png) [@bakman2](https://discourse.nodered.org/u/bakman2)\
**Post date:** [17 March 2020 17:30 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/6 "2020-03-17T17:30:08Z")

</div>

I initially got distracted from the issue indeed, but I think the problem is that the flow variable does not exist when injecting into the subflow. See bottom of my response.

---

<div class="post-metadata">

**Author:** ![nileio](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/nileio/32/9444_2.png) [@nileio](https://discourse.nodered.org/u/nileio)\
**Post date:** [17 March 2020 21:21 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/7 "2020-03-17T21:21:50Z")

</div>

@bakman2 thanks for suggestion. I use `$parent.somecontext` in the correct way. Both the suggested ways using either `Change` node and Jsonata in my flow fail the same way. the weird thing is that I dont even get `undefined` which should be the case when for any reason the value can't be retrieved by the node. what really happens is that I get nothing at all from the output port..  
I will upload a video showing the issue with my subflows

---

<div class="post-metadata">

**Author:** ![nileio](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/nileio/32/9444_2.png) [@nileio](https://discourse.nodered.org/u/nileio)\
**Post date:** [17 March 2020 21:28 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/8 "2020-03-17T21:28:21Z")

</div>

> [@bakman2](#):
>
> but I think the problem is that the flow variable does not exist when injecting into the subflow

I wish it was just that. The variable definitely exists and initialised. I can use it from the root flow with no issue. Note : this issue is little hard to reproduce because it does not happen when you create a subflow from scratch, it happens when you either create a subflow from existing nodes or when you export a flow and import it somewhere else when your subflow has lots of other nodes . as I said in my original post, I am unable to find out what is the exact way to reproduce it, however, it is happening in my environments right now ..

---

<div class="post-metadata">

**Author:** ![nileio](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/nileio/32/9444_2.png) [@nileio](https://discourse.nodered.org/u/nileio)\
**Post date:** [17 March 2020 22:03 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/9 "2020-03-17T22:03:43Z")

</div>

Hi @knolleary and @bakman2 , here is a video demonstrating the problem to the best I could..  
notice how the "problem subflow" does not produce any output at all. I took all my other nodes in the problem subflow off-screen to focus on the issue..  
let me know your thoughts..

 ![2020-03-18_09-16-47 (1)](https://us1.discourse-cdn.com/flex026/uploads/nodered/original/3X/f/8/f8126dc099b8c436785f7e2d28714d9bd596805a.gif)

---

<div class="post-metadata">

**Author:** ![nileio](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/nileio/32/9444_2.png) [@nileio](https://discourse.nodered.org/u/nileio)\
**Post date:** [19 March 2020 00:51 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/10 "2020-03-19T00:51:19Z")

</div>

hi @knolleary can you confirm if I need to create an issue on github for this ? I had to re-build all subflows manually one by one to get around it for now so keen to log it for a fix if possible or find out the root cause..

thanks!

---

<div class="post-metadata">

**Author:** ![nileio](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/nileio/32/9444_2.png) [@nileio](https://discourse.nodered.org/u/nileio)\
**Post date:** [24 March 2020 22:49 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/11 "2020-03-24T22:49:49Z")

</div>

update on this issue can be found at [https://github.com/node-red/node-red/issues/2513](https://github.com/node-red/node-red/issues/2513) & [https://github.com/drmibell/node-red-contrib-queue-gate/issues/10](https://github.com/drmibell/node-red-contrib-queue-gate/issues/10)

---

<div class="post-metadata">

**Author:** ![drmibell](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/drmibell/32/8424_2.png) [@drmibell](https://discourse.nodered.org/u/drmibell)\
**Post date:** [27 March 2020 20:30 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/12 "2020-03-27T20:30:35Z")

</div>

> [@knolleary](#):
>
> If you are able to narrow it down at all, or at very least give an example sequence of steps that does reliably reproduce it, that would be very helpful.

I think I can reproduce this issue using just one line of JS code, but I still do not understand the real cause. These are the steps:

1. Install `node-red-contrib-lower`. (Almost any other node could be used, but this one is easy.)
2. Import this flow (based on the one posted by @nileio):

```auto
[{"id":"1c9112c4.636bfd","type":"subflow","name":"subflow","info":"","category":"","in":[{"x":220,"y":80,"wires":[{"id":"a73a7eda.92af5"}]}],"out":[{"x":510,"y":80,"wires":[{"id":"a73a7eda.92af5","port":0}]}],"env":[],"color":"#DEB887"},{"id":"a73a7eda.92af5","type":"change","z":"1c9112c4.636bfd","name":"","rules":[{"t":"set","p":"payload","pt":"msg","to":"$parent.somecontext","tot":"flow"}],"action":"","property":"","from":"","to":"","reg":false,"x":380,"y":80,"wires":[[]]},{"id":"f35176ad.2aeec","type":"lower-case","z":"1c9112c4.636bfd","name":"","x":380,"y":140,"wires":[[]]},{"id":"b105ba0c.0e5f","type":"inject","z":"275b6d65.ead67a","name":"inject #1","topic":"","payload":"","payloadType":"date","repeat":"","crontab":"","once":false,"onceDelay":0.1,"x":180,"y":80,"wires":[["75695b82.5772ac"]]},{"id":"75695b82.5772ac","type":"change","z":"275b6d65.ead67a","name":"","rules":[{"t":"set","p":"somecontext","pt":"flow","to":"test","tot":"str"}],"action":"","property":"","from":"","to":"","reg":false,"x":370,"y":80,"wires":[[]]},{"id":"50266c7d.cfbd84","type":"subflow:1c9112c4.636bfd","z":"275b6d65.ead67a","name":"","env":[],"x":385,"y":145,"wires":[["6a4563f.1df9e9c"]]},{"id":"17bdb382.6c814c","type":"inject","z":"275b6d65.ead67a","name":"inject #2","topic":"","payload":"","payloadType":"date","repeat":"","crontab":"","once":false,"onceDelay":0.1,"x":235,"y":145,"wires":[["50266c7d.cfbd84"]]},{"id":"6a4563f.1df9e9c","type":"debug","z":"275b6d65.ead67a","name":"","active":true,"tosidebar":true,"console":false,"tostatus":false,"complete":"false","x":545,"y":145,"wires":[]}]

```

1. At this point, triggering `inject #2` should give _undefined_ as the `debug` output. If you trigger `inject #1`, followed by `inject #2`, the debug output should be "test" and continue that way for future triggers of `inject #2`. (All this is to be expected.)
2. Edit `lowercase.js` (generally found in ~/.node-red/node\_modules/node-red-contrib-lower/lower/) by adding one line (the 5th):

```auto
module.exports = function(RED) {
    function LowerCaseNode(config) {
        RED.nodes.createNode(this,config);
        var node = this;
        var context = node.context();
        this.on('input', function(msg) {
            msg.payload = msg.payload.toLowerCase();
            node.send(msg);
        });
    }
    RED.nodes.registerType("lower-case",LowerCaseNode);
}

```

1. Repeat step 3. There should be no change.
2. Stop and restart NR.
3. Triggering `inject #2` now produces no output from the subflow.

Once you have reached step 6, the only way I have found to restore normal behavior is to edit the subflow template to remove the `lower-case` node **AND** restart NR.

I suspect that an instance of a node inside a subflow does not really have its own node context, at least not in the usual way, but I've reached the limit of my understanding/guesswork.

---

<div class="post-metadata">

**Author:** ![knolleary](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/knolleary/32/3_2.png) [@knolleary](https://discourse.nodered.org/u/knolleary)\
**Post date:** [27 March 2020 20:34 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/13 "2020-03-27T20:34:54Z")

</div>

Thanks @drmibell

I updated the [issue this morning](https://github.com/node-red/node-red/issues/2513#issuecomment-604936447).

> The issue is all of the nodes you've identified access their context in their constructor function. They are absolutely allowed to do that, but it exposes a problem with the way subflow instance context is created - it gets done _after_ the nodes are created. But as the nodes have already accessed their context by that time, it breaks the link between the node context and its parent.

I've got a fix working, but it needs lots of testing as it has changed some of the core logic around context handling.

---

<div class="post-metadata">

**Author:** ![drmibell](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/drmibell/32/8424_2.png) [@drmibell](https://discourse.nodered.org/u/drmibell)\
**Post date:** [27 March 2020 20:56 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/14 "2020-03-27T20:56:27Z")

</div>

Thanks, Nick. So my guess was "close but no cigar."

Should I put notes in the documentation of my nodes to warn against using them in subflows until the fix is available?

---

<div class="post-metadata">

**Author:** ![drmibell](https://sea2.discourse-cdn.com/flex026/user_avatar/discourse.nodered.org/drmibell/32/8424_2.png) [@drmibell](https://discourse.nodered.org/u/drmibell)\
**Post date:** [9 April 2020 18:06 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/15 "2020-04-09T18:06:06Z")

</div>

Just for the record, I believe this has been fixed in NR version 1.0.5 🙂

---

<div class="post-metadata">

**Author:** ![system](https://us1.discourse-cdn.com/flex026/uploads/nodered/original/1X/d073cd938eafa2e558d7c2cd59003b3ef4963033.png) [@system](https://discourse.nodered.org/u/system)\
**Post date:** [8 June 2020 18:08 UTC](https://discourse.nodered.org/t/issues-using-parent-from-a-subflow/23231/16 "2020-06-08T18:08:24Z")

</div>

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.