FelixL
April 18, 2023, 4:02pm
1
Sadly, my Laptop only has 16 GB ram right now (soldered)
I am using Firefox right now, and itβs the only program opened on my Laptop with only 1 Tab of N8N with a quite big workflow.
After half an hour working on the workflow, Firefox uses over 8 GB ram.
Reopening the workflow helps.
Is that expected behaviour right now?
If not, is there any info/logs I can provide to help to get to the root cause of this?
Information on your n8n setup
n8n version: 0.222.3
Database (default: SQLite): MySQL
n8n EXECUTIONS_PROCESS setting (default: own, main):
Running n8n via (Docker, npm, n8n cloud, desktop app): Docker
Operating system: Ubuntu (Windows 11 on Client)
Jon
April 19, 2023, 7:10am
2
Hey @FelixL ,
I had noticed a similar issue last night while working on a workflow, I managed to create a workflow that triggers the same issue which is handy.
Would you be able to also share the workflow you were working on so we can check for any patterns just to confirm we are on the right path?
FelixL
April 19, 2023, 7:30am
3
Yeah sure, but I do not know any more which workflow I worked on.
Iβll work a few hours today with N8N, as soon as I see similar behaviour again Iβll send you a pm (or here, depending on the workflow) with the workflow.
1 Like
FelixL
April 19, 2023, 9:37am
4
I couldnβt reproduce the issue right now, Iβll make an update when I have the problem again.
Should I try your workflow on my system?
Which nodes/combination of nodes are you using? (To narrow down which workflow caused this in my case.)
Jon
April 19, 2023, 10:25am
5
Hey @FelixL ,
For me it was my workflow that clears old workflows out, Specifically though it appears to be when using set nodes with lots of data coming in. We have the team looking into it though so I didnβt want to lead you down any specific paths with it
FelixL
April 19, 2023, 10:45am
6
ok, got a few set nodes with a lot of items ^^
I am using the set nodes to remove not needed fields to help with memory requirements. I hope thatβs the right way of thinking here.
Iβve also found the workflow which used 8GB of RAM on the client yesterday:
It has two pipedrive nodes which read about 600 deals at the same time and processes the data for our dashboard. (I have to split it two workflows, but thats at the moment not really easy todo: Bug node selection does not work )
When execution the workflow it uses 800MB on the Client (150MB on the Server), but as soon as the Execution finishes it releases that memory, which was not the case yesterday.
Writing another post rn, where one of my workflows got a memory problem and with it failed multiple other workflows.
FelixL
April 19, 2023, 10:52am
7
Another workflow which also caused problems yesterday:
Workflow
This workflow has 3 pipedrive input nodes, one with about 7000 items, one with, 600 items and one with 1300 items. all 3 inputs get cleaned with a set node.
Iβve already split that workflow into 4 smaller workflows.
Jon
April 19, 2023, 11:03am
8
That is perfect, So I suspect it is the same thing. Thanks for sharing those.
FelixL
April 19, 2023, 12:55pm
9
If there is anything else I can provide, i am happy to assist (:
Here is a memory report of one of the workflows:
made in Firefox with βabout:memoryβ
memory report
webIsolated=https://example.at (pid 25192)
Explicit Allocations
1,934.90 MB (100.0%) -- explicit
βββ1,721.66 MB (88.98%) -- window-objects
β βββββ591.36 MB (30.56%) -- top(https://n8n.example.at/workflow/19/executions/63421, id=8589934599)
β β βββ404.36 MB (20.90%) -- active
β β β βββ212.51 MB (10.98%) -- window(https://n8n.example.at/workflows/demo)
β β β β βββ206.72 MB (10.68%) -- js-realm(https://n8n.example.at/workflows/demo, about:blank)
β β β β β βββ200.30 MB (10.35%) -- classes
β β β β β β ββββ58.32 MB (03.01%) -- class(Object)/objects
β β β β β β β βββ41.59 MB (02.15%) ββ gc-heap
β β β β β β β βββ16.73 MB (00.86%) ++ malloc-heap
β β β β β β ββββ57.05 MB (02.95%) -- class(Array)/objects
β β β β β β β βββ44.07 MB (02.28%) ββ gc-heap
β β β β β β β βββ12.98 MB (00.67%) ++ malloc-heap
β β β β β β ββββ34.07 MB (01.76%) -- class(Function)/objects
β β β β β β β βββ33.72 MB (01.74%) ββ gc-heap
β β β β β β β ββββ0.35 MB (00.02%) ββ malloc-heap/slots
β β β β β β ββββ31.20 MB (01.61%) -- class(Call)/objects
β β β β β β β βββ31.15 MB (01.61%) ββ gc-heap
β β β β β β β ββββ0.05 MB (00.00%) ββ malloc-heap/slots
β β β β β β ββββ19.68 MB (01.02%) ++ (10 tiny)
β β β β β βββββ6.41 MB (00.33%) ++ (5 tiny)
β β β β βββββ5.80 MB (00.30%) ++ (3 tiny)
β β β βββ191.84 MB (09.91%) -- window(https://n8n.example.at/workflow/19/executions/63421)
β β β βββ185.17 MB (09.57%) -- js-realm(https://n8n.example.at/workflow/19/executions/62662, about:blank)
β β β β βββ177.45 MB (09.17%) -- classes
β β β β β ββββ54.14 MB (02.80%) -- class(Object)/objects
β β β β β β βββ38.43 MB (01.99%) ββ gc-heap
β β β β β β βββ15.71 MB (00.81%) ++ malloc-heap
β β β β β ββββ50.22 MB (02.60%) -- class(Array)/objects
β β β β β β βββ40.09 MB (02.07%) ββ gc-heap
β β β β β β βββ10.13 MB (00.52%) ++ malloc-heap
β β β β β ββββ30.58 MB (01.58%) -- class(Function)/objects
β β β β β β βββ30.25 MB (01.56%) ββ gc-heap
β β β β β β ββββ0.33 MB (00.02%) ββ malloc-heap/slots
β β β β β ββββ27.47 MB (01.42%) -- class(Call)/objects
β β β β β β βββ27.42 MB (01.42%) ββ gc-heap
β β β β β β ββββ0.05 MB (00.00%) ββ malloc-heap/slots
β β β β β ββββ15.04 MB (00.78%) ++ (10 tiny)
β β β β βββββ7.72 MB (00.40%) ++ (6 tiny)
β β β βββββ6.67 MB (00.34%) ++ (3 tiny)
β β βββ187.01 MB (09.66%) -- js-zone(0x1e59e4d8300)
β β ββββ64.30 MB (03.32%) ++ (15 tiny)
β β ββββ61.04 MB (03.15%) -- property-maps
β β β βββ33.89 MB (01.75%) -- gc-heap
β β β β βββ31.41 MB (01.62%) ββ dict
β β β β ββββ2.47 MB (00.13%) ++ (2 tiny)
β β β βββ27.16 MB (01.40%) -- malloc-heap
β β β βββ26.93 MB (01.39%) ββ tables
β β β ββββ0.23 MB (00.01%) ββ children
β β ββββ41.66 MB (02.15%) ββ unused-gc-things
β β ββββ20.00 MB (01.03%) ββ unique-id-map
β βββββ430.08 MB (22.23%) -- top(none)/detached/window(https://n8n.example.at/workflows/demo)
β β βββ420.39 MB (21.73%) -- js-realm(https://n8n.example.at/workflows/demo, about:blank)
β β β βββ408.04 MB (21.09%) -- classes
β β β β βββ119.29 MB (06.17%) -- class(Object)/objects
β β β β β ββββ84.22 MB (04.35%) ββ gc-heap [2]
β β β β β ββββ35.07 MB (01.81%) -- malloc-heap
β β β β β βββ33.96 MB (01.76%) ββ slots [2]
β β β β β ββββ1.11 MB (00.06%) ββ elements/normal [2]
β β β β βββ115.00 MB (05.94%) -- class(Array)/objects
β β β β β ββββ89.31 MB (04.62%) ββ gc-heap [2]
β β β β β ββββ25.69 MB (01.33%) -- malloc-heap
β β β β β βββ24.70 MB (01.28%) ββ elements/normal [2]
β β β β β ββββ0.99 MB (00.05%) ββ slots [2]
β β β β ββββ68.99 MB (03.57%) -- class(Function)/objects
β β β β β βββ68.22 MB (03.53%) ββ gc-heap [2]
β β β β β ββββ0.77 MB (00.04%) ββ malloc-heap/slots [2]
β β β β ββββ63.01 MB (03.26%) -- class(Call)/objects
β β β β β βββ62.92 MB (03.25%) ββ gc-heap [2]
β β β β β ββββ0.10 MB (00.01%) ββ malloc-heap/slots [2]
β β β β ββββ38.47 MB (01.99%) -- class(Set)/objects
β β β β β βββ37.30 MB (01.93%) ββ malloc-heap/misc [2]
β β β β β ββββ1.17 MB (00.06%) ββ gc-heap [2]
β β β β βββββ3.28 MB (00.17%) ++ (10 tiny)
β β β ββββ12.35 MB (00.64%) ++ (3 tiny)
β β βββββ9.68 MB (00.50%) ++ (3 tiny)
β βββββ373.20 MB (19.29%) -- top(https://n8n.example.at/executions, id=8589934594)
β β βββ287.77 MB (14.87%) -- active/window(https://n8n.example.at/executions)
β β β βββ269.62 MB (13.93%) -- js-realm(https://n8n.example.at/workflow/16/executions/62660, about:blank)
β β β β βββ260.89 MB (13.48%) -- classes
β β β β β βββ104.60 MB (05.41%) -- class(Object)/objects
β β β β β β ββββ57.54 MB (02.97%) ββ gc-heap
β β β β β β ββββ47.06 MB (02.43%) -- malloc-heap
β β β β β β βββ46.46 MB (02.40%) ββ slots
β β β β β β ββββ0.60 MB (00.03%) ββ elements/normal
β β β β β ββββ62.24 MB (03.22%) -- class(Array)/objects
β β β β β β βββ51.85 MB (02.68%) ββ gc-heap
β β β β β β βββ10.40 MB (00.54%) ++ malloc-heap
β β β β β ββββ38.32 MB (01.98%) -- class(Function)/objects
β β β β β β βββ37.45 MB (01.94%) ββ gc-heap
β β β β β β ββββ0.87 MB (00.05%) ββ malloc-heap/slots
β β β β β ββββ33.91 MB (01.75%) -- class(Call)/objects
β β β β β β βββ33.87 MB (01.75%) ββ gc-heap
β β β β β β ββββ0.05 MB (00.00%) ββ malloc-heap/slots
β β β β β ββββ21.81 MB (01.13%) ++ (21 tiny)
β β β β βββββ8.73 MB (00.45%) ++ (6 tiny)
β β β ββββ18.15 MB (00.94%) ++ (3 tiny)
β β ββββ85.43 MB (04.42%) -- js-zone(0x1e599583100)
β β βββ58.91 MB (03.04%) ++ (17 tiny)
β β βββ26.52 MB (01.37%) -- property-maps
β β βββ24.49 MB (01.27%) -- gc-heap
β β β βββ21.95 MB (01.13%) ββ dict
β β β ββββ2.53 MB (00.13%) ++ (2 tiny)
β β ββββ2.03 MB (00.11%) ++ malloc-heap
β βββββ327.01 MB (16.90%) -- top(https://n8n.example.at/executions, id=53)
β βββ279.40 MB (14.44%) -- active/window(https://n8n.example.at/executions)
β β βββ263.73 MB (13.63%) -- js-realm(https://n8n.example.at/)
β β β βββ255.08 MB (13.18%) -- classes
β β β β ββββ97.53 MB (05.04%) -- class(Object)/objects
β β β β β βββ55.60 MB (02.87%) ββ gc-heap
β β β β β βββ41.92 MB (02.17%) -- malloc-heap
β β β β β βββ41.31 MB (02.13%) ββ slots
β β β β β ββββ0.62 MB (00.03%) ββ elements/normal
β β β β ββββ62.53 MB (03.23%) -- class(Array)/objects
β β β β β βββ51.61 MB (02.67%) ββ gc-heap
β β β β β βββ10.93 MB (00.56%) ++ malloc-heap
β β β β ββββ38.62 MB (02.00%) -- class(Function)/objects
β β β β β βββ37.82 MB (01.95%) ββ gc-heap
β β β β β ββββ0.80 MB (00.04%) ββ malloc-heap/slots
β β β β ββββ34.47 MB (01.78%) -- class(Call)/objects
β β β β β βββ34.42 MB (01.78%) ββ gc-heap
β β β β β ββββ0.05 MB (00.00%) ββ malloc-heap/slots
β β β β ββββ21.94 MB (01.13%) ++ (20 tiny)
β β β βββββ8.65 MB (00.45%) ++ (6 tiny)
β β ββββ15.67 MB (00.81%) ++ (3 tiny)
β ββββ47.61 MB (02.46%) ++ js-zone(0x1e58b750100)
ββββββ74.07 MB (03.83%) -- js-non-window
β βββ29.84 MB (01.54%) -- gc-heap
β β βββ22.63 MB (01.17%) ββ chunk-admin
β β ββββ7.21 MB (00.37%) ++ (2 tiny)
β βββ25.25 MB (01.30%) ++ runtime
β βββ18.98 MB (00.98%) ++ (2 tiny)
ββββββ68.87 MB (03.56%) ββ heap-unclassified
ββββββ40.91 MB (02.11%) -- heap-overhead
β βββ26.68 MB (01.38%) ++ bin-unused
β βββ14.23 MB (00.74%) ++ (2 tiny)
ββββββ29.39 MB (01.52%) ++ (22 tiny)
netroy
April 20, 2023, 4:03pm
10
@FelixL A potentially related memory leak was fixed in this PR and released in 0.224.2
.
Can you please upgrade, and if that fixes the issue for you, please let is know ?
2 Likes
FelixL
April 22, 2023, 2:36pm
11
@netroy Iβve just installed it, Iβll give some feedback as soon as Iβve worked a few hours with n8n.
1 Like
FelixL
April 27, 2023, 10:23am
12
Hi @netroy ,
I have not had any excessive ram usage in the last few days with any of my workflows.
But I just noticed, when I open this forum post, Firefox uses about 1.2 GB more RAM.
As soon as I go to another page on the N8N forum, 1.2 GB get released again.
netroy
April 27, 2023, 1:19pm
13
Thanks for reporting. That issue should be fixed now as well.