Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Pane is always loading #787

Open
notuntoward opened this issue Sep 16, 2024 · 38 comments
Open

Pane is always loading #787

notuntoward opened this issue Sep 16, 2024 · 38 comments
Labels
help wanted Extra attention is needed

Comments

@notuntoward
Copy link

notuntoward commented Sep 16, 2024

Lately, Smart Connections has stopped working. When I start Obsidian, I see the plugin embedding messages, which as far as I can tell, finish successfully. But when I open the smart connections pane, I don't see related note links. Instead, all I ever I see is this:

image

@christianngarcia christianngarcia added the help wanted Extra attention is needed label Sep 16, 2024
@dmdamen
Copy link

dmdamen commented Sep 16, 2024

Same here. Using Obsidian on Fedora 40. Installed using flatpak.

Fresh Obsidian install, added a few notes. No other community plugins installed.

@ductiletoaster
Copy link

I can confirm this happened immediately after upgrading to 2.2.70.

System Info

  • Windows 11 Pro
  • CPU AMD 3800x
  • RAM 32GB
  • GPU 4070 TI Super

My fault it setup with Obsidian Sync and End-to-End encryption. Previously I was running 2.1.101 when this worked. I am in fact still running that version on my work laptop (Macbook Pro M3) and it still works well.

@brianpetro
Copy link
Owner

Are there any errors in the console logs?

Accessing the Developer Console:
MacOS: Cmd+Option+i
Windows: Ctrl+SHIFT+i
Or in the menu, View → Toggle Developer Tools.
And in those developer tools, the tab “Console”.

@Postur
Copy link

Postur commented Sep 17, 2024

Got the same issue after updating to 2.2.70.

My console log errors and warnings:

2024-09-17 17:48:43.249300 [W:onnxruntime:, session_state.cc:1168 VerifyEachNodeIsAssignedToAnEp] Some nodes 
were not assigned to the preferred execution providers which may or may not have an negative impact on performance. 
e.g. ORT explicitly assigns shape related ops to CPU to improve perf
2024-09-17 17:48:43.249800 [W:onnxruntime:, session_state.cc:1170 VerifyEachNodeIsAssignedToAnEp] Rerunning with
 verbose output on a non-minimal build will show node assignments.
DevTools failed to load source map: Could not load content for 
https://cdn.jsdelivr.net/npm/@huggingface/transformers.min.js.map: HTTP error: status code 404, 
net::ERR_HTTP_RESPONSE_CODE_FAILURE

@brianpetro
Copy link
Owner

That shouldn't be problematic

Anything else?

🌴

@Postur
Copy link

Postur commented Sep 17, 2024

image

This is about it

running lsof -i :37042 shows obsidian is running on port 37042, which I assume is correct.

@BenjaminCarpenter480
Copy link

I have also experenced this error, the developer log seems to have the following warnings and errors, run on ubuntu 24.04:

app.js:1 Obsidian Developer Console
VM197 plugin:smart-connections:13356 Loading Smart Connections v2...
VM197 plugin:smart-connections:13372 loading env
about:srcdoc:309 init
about:srcdoc:312 load Object
about:srcdoc:135 Object
about:srcdoc:239 [Transformers] Using GPU
[email protected]:100 WebGPU is experimental on this platform. See https://github.com/gpuweb/gpuweb/wiki/Implementation-Status#implementation-status
[email protected]:100 Failed to create WebGPU Context Provider
Du @ [email protected]:100
sd @ [email protected]:100
init @ [email protected]:100
about:srcdoc:157 Error loading model TaylorAI/bge-micro-v2: Error: no available backend found. ERR: [webgpu] Error: Failed to get GPU adapter. You may need to enable flag "--enable-unsafe-webgpu" if you are using Chrome.
    at l ([email protected]:100:1798)
    at async e.create ([email protected]:100:18017)
    at async f ([email protected]:151:1428)
    at async [email protected]:175:15503
    at async Promise.all (index 0)
    at async P ([email protected]:175:13530)
    at async Promise.all (index 0)
    at async U.from_pretrained ([email protected]:175:22668)
    at async Fo.from_pretrained ([email protected]:175:57231)
    at async Promise.all (index 1)
load @ about:srcdoc:157
VM197 plugin:smart-connections:9355 model loaded
VM197 plugin:smart-connections:7063 SmartEmbedModel already loaded for TaylorAI/bge-micro-v2
VM197 plugin:smart-connections:13393 env loaded
VM197 plugin:smart-connections:13374 Smart Connections v2 loaded

@ravn-gt
Copy link

ravn-gt commented Sep 17, 2024

I too am having infinite loading problems. There is no separate error displayed in the developer console.
The problem may have occurred when the model settings were integrated into the smart chat settings tab in version 2.2.7.
I was using a local LLM.

image

@EasternPA
Copy link

image
I'm also experiencing it, but my errors from running the AppImage appear unrelated at first glance. These may be helpful. Ubuntu 22.04.4.

@brianpetro
Copy link
Owner

@Postur the port is for a few actions that the Smart Connect app depends on (primarily rendering dataviews), but shouldn't have any impact on the Smart Connections plugin, whether it "connects" or not.

@BenjaminCarpenter480 try using the old version of transformers, see #749 (comment)

@ravn-gt messages seem to disappear when I use "Based on my notes" queries with a local model. Is that what you're experiencing?

@EasternPA none of those errors are related to Smart Connections

🌴

@BenjaminCarpenter480
Copy link

BenjaminCarpenter480 commented Sep 17, 2024

Following your advice I turned on legacy transformers, restarted obsidian and noted the same behaviour and (as far as a quick glance suggested) error logs, see attached:

WithLegacyOn.log

Also included for reference the logs with legacy off
WithLegacyOff.log

Hope this is of some use in identifying an issue - if it is the same one others have found

@PhotoVisionProject
Copy link

Unfortunately, I encountered the same issue right after updating to 2.2.70—endless loading times. Enabling legacy transformers didn’t help either.

@brianpetro
Copy link
Owner

@BenjaminCarpenter480 try restarting after turning the legacy option on (in the logs it looks like it is still trying to use the new version) 🌴

@BenjaminCarpenter480
Copy link

I set the legacy transformer option to on and did a full restart of my PC. I still get the Transformers Using GPU message in the logs?

image

@brianpetro
Copy link
Owner

@BenjaminCarpenter480 thanks for letting me know.

The setting was broken. The bug is fixed in v2.2.71.

After updating, you may need to re-toggle the setting to get it to take effect. When it does, you should see something like this in the console

Screenshot 2024-09-17 at 5 59 16 PM

Those errors are OK and the embedding should work after that
🌴

@ravn-gt
Copy link

ravn-gt commented Sep 18, 2024

After the 2.2.72 update, the issue of only “Loading Smart Connections....” appearing in the pane was resolved.
It works fine now!

@brianpetro, your plugin is a game changer. Interesting and useful. Thank you for your hard work.

@oscarvdrgrgrgr
Copy link

@brianpetro
in update 2.72 it works fine but only if you decide to embed your notes with the TailorAI option when embedding with other options like text embedding 3 small the infinite loading problem continues

@brianpetro
Copy link
Owner

@oscarvdrgrgrgr are there any errors in the console? I just embedded a full vault with OpenAI text-small, seems to be working 🌴

@oscarvdrgrgrgr
Copy link

oscarvdrgrgrgr commented Sep 18, 2024

image
@brianpetro

@nweii
Copy link

nweii commented Sep 18, 2024

I'm getting a similar-looking error to @oscarvdrgrgrgr and stuck on "Loading Smart Connections…" in my sidebar. I had existing embeddings using text-embedding-3-small-512. I've already tried switching Legacy/modern transformers to no avail. I'm on the latest version 2.2.73.
Screenshot 2024-09-18 at 3 14 56 PM

@BenjaminCarpenter480
Copy link

@BenjaminCarpenter480 thanks for letting me know.

The setting was broken. The bug is fixed in v2.2.71.

After updating, you may need to re-toggle the setting to get it to take effect. When it does, you should see something like this in the console

Screenshot 2024-09-17 at 5 59 16 PM

Those errors are OK and the embedding should work after that
🌴

@brianpetro Just to note that this now all works as expected with the legacy transformers cheers!

@brianpetro
Copy link
Owner

@nweii working on a fix 🌴

brianpetro added a commit to brianpetro/jsbrains that referenced this issue Sep 18, 2024
@brianpetro
Copy link
Owner

@nweii should be fixed in 2.2.74 🌴

@ArthurSrz
Copy link

I updated to the latest version 2.2.76, on windows and still get the issues
Capture d'écran 2024-09-19 170952

I did try to toggle on and off the legacy transformers settings but it didn't work. Is there another fix shortcut I didn't try ?

@brianpetro
Copy link
Owner

@ArthurSrz that looks like a different issue. Likely from an unrecognized file type 🌴

@notuntoward
Copy link
Author

notuntoward commented Sep 20, 2024

I toggled off Legacy Transformers (no GPU), and then back on. Then, the pane looked like this, with no links displayed:

image

The links finally did appear after I clicked the "recycle" button: much better.

But when there's a state change, should the plugin "click the recycle button" by itself?

@uwe999
Copy link

uwe999 commented Sep 20, 2024

I installed the plugin today for the first time and got stuck on the "loading pane" as well. Toggling the "Legacy transformer" switch started the "making smart connections" process

@sh4d0wl3ss
Copy link

I'm getting the same thing. In case it's useful here's what's on my console:

Screenshot 2024-09-24 170802

@sh4d0wl3ss
Copy link

No difference with Legacy transformer on or off.

@huachuman
Copy link

image

@3e9b
Copy link

3e9b commented Sep 26, 2024

Screenshot 2024-09-26 at 1 00 19 PM

@brianpetro
Copy link
Owner

brianpetro commented Sep 26, 2024

@sh4d0wl3ss good to know

@huachuman which model is that happening with?

@3e9b all of those errors are OK except for the one I responded to in the other issue with the same image #797 (comment)

🌴

@notuntoward
Copy link
Author

notuntoward commented Oct 3, 2024

Smart Connections was working for me, but I just installed 2.2.77 and it's now stuck again at "loading smart connections."

The popup notifying me of this smart connections update mentioned something about two steps refresh, but I can't find the relevant smart connections commands or settings.

@brianpetro
Copy link
Owner

@notuntoward
Copy link
Author

@notuntoward see the image in the release notes here https://github.com/brianpetro/obsidian-smart-connections/releases/tag/2.2.77 🌴

Thanks. Is that image what I should see if I click the gear icon when smart connections is running? If so, I can't see it because the plugin freezes at this stage:

image

@brianpetro
Copy link
Owner

@notuntoward in the obsidian main settings 🌴

Screenshot 2024-10-03 at 10 50 06 PM

@cristianvasquez
Copy link

I have also experenced this error, the developer log seems to have the following warnings and errors, run on ubuntu 24.04:

app.js:1 Obsidian Developer Console
VM197 plugin:smart-connections:13356 Loading Smart Connections v2...
VM197 plugin:smart-connections:13372 loading env
about:srcdoc:309 init
about:srcdoc:312 load Object
about:srcdoc:135 Object
about:srcdoc:239 [Transformers] Using GPU
[email protected]:100 WebGPU is experimental on this platform. See https://github.com/gpuweb/gpuweb/wiki/Implementation-Status#implementation-status
[email protected]:100 Failed to create WebGPU Context Provider
Du @ [email protected]:100
sd @ [email protected]:100
init @ [email protected]:100
about:srcdoc:157 Error loading model TaylorAI/bge-micro-v2: Error: no available backend found. ERR: [webgpu] Error: Failed to get GPU adapter. You may need to enable flag "--enable-unsafe-webgpu" if you are using Chrome.
    at l ([email protected]:100:1798)
    at async e.create ([email protected]:100:18017)
    at async f ([email protected]:151:1428)
    at async [email protected]:175:15503
    at async Promise.all (index 0)
    at async P ([email protected]:175:13530)
    at async Promise.all (index 0)
    at async U.from_pretrained ([email protected]:175:22668)
    at async Fo.from_pretrained ([email protected]:175:57231)
    at async Promise.all (index 1)
load @ about:srcdoc:157
VM197 plugin:smart-connections:9355 model loaded
VM197 plugin:smart-connections:7063 SmartEmbedModel already loaded for TaylorAI/bge-micro-v2
VM197 plugin:smart-connections:13393 env loaded
VM197 plugin:smart-connections:13374 Smart Connections v2 loaded

Hi, I got the same error, I ran Obsidian as follows:

./Obsidian.AppImage  --enable-unsafe-webgpu --enable-features=Vulkan

And it began indexing using GPU

@brianpetro
Copy link
Owner

Great to know! Thanks for contributing that solution @cristianvasquez 😊🌴

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
help wanted Extra attention is needed
Projects
None yet
Development

No branches or pull requests