Glances 4.X breaks support with "gethomepage.dev" #2763
Comments
|
Same here |
|
In services.yaml, adding 'version: 4' to each widget should solve your problem. Details can be found in the homepage documentation for glances regarding the support for glances 4.X. |
|
There is still an issue that I am discussing with the Hompage dev. Occasionally the sensors module outputs incorrect data and Homepage throws an error. It works for a few seconds but then the error is thrown. Accessing the API endpoint via curl even with the error seems to send back correct data. |
|
OK so testing with curl, I have on a couple of occasions had the API return |
|
looking at the web ui, the whole sensors area just drops out and there's a spike of cpu while it's loading. I'll revert to the last v3 docker image for now Screen.Recording.2024-05-13.at.8.37.48.AM.mov |
|
Homepage dev here. We’ve had v4 support ready to go long in anticipation of this. I think the sensors issue is separate entirely. I would suggest that this issue can be closed. |
|
@shamoon |
|
Sensor information is not showing at all in v4 of Glances, so Homepage has nothing to reference to for the Temperature widget. I don't see how this is a Homepage issue. |
|
It's the API,. occasionally it returns blank data for sensor. I haven't had an issue with the data endpoints. It's not really a Homepage issue, other than it not handling the blank API return, but the CPU spikes and blank data shouldn't really happen to begin with. |
|
I went back to v3.4 for now anyway. |
|
Concerning the sensors issue, can ou please test it regarding to the comment #2765 (comment). For the others requests, check the new API here: https://github.com/nicolargo/glances/blob/develop/docs/api.rst |
|
Addding version: 4 In the widgets.yaml works for the header widget only...
|
- Glances Viz v4:
- CPU Usage:
widget:
type: glances
url: http://localhost:61208
metric: cpu
version: 4
- Memory:
widget:
type: glances
url: http://localhost:61208
metric: memory
version: 4 |
Are you trying to point out it should work with v4? To show again, this is my .yaml snippet: But it looks like this:
|
I have run the curl command on my instance: it sometimes loses the infos, but i can not relate if this is relevant directly with my issue, since i dont get any infos. |
|
Again, theres more than one issue discussed here which is confusing. But regarding sensors, I can make homepage handle |
|
I have never seen it happen on v3. |
Yeah that works, Thank you! |
In regards to the empty array, the sensors data which contains the below items:
Hence the sensors API returning empty array is a possible and valid case in some scenarios. So I think integrations should these cases unless their usecase doesn't need to account for the edge-cases specified above. Please try out glances on other devices/platforms and i believe you can notice the differences I've mentioned above. |
|
Sorry, I mean intermittently returning the empty array. If a sensor just isnt supported at all I agree, but like the comment above #2763 (comment) where it returns expected data on some API calls and not on others. If the merged PR you linked to prevents that then all good, thanks! |
|
@joshoram80 |
Intermittently missing entries or returning empty array was a bug due to some race conditions. @nicolargo seems to be making the docker images releases now for dev. Thanks for the speedy release! Will close the issue in next couple days as long as no other issues arise. |
|
@RazCrimson all seems to be working fine now ;D. No errors in homepage or with curl |
|
Something is still wrong it seems... I came back home after being away for a few hours and the error is still there. Same as before. I can't see exactly what caused it, whether it was an empty API call return or not, but it only takes one for the error to happen. While i was actively monitoring it it appeared to be working correctly. |
|
I am not sure what's the behaviour of the code used in the Graphs. But FYI, a sensor can disappear from the sensors list or a newer sensor can appear in the list while glances is running. Reason for the above is glances merely tries to fetch all sensor data and sends them. So any activity like adding/removing drives or fans can cause sensor changes. Handling the above case and empty array should hopefully fix all issues on homepage end. |
|
I can look at it on the homepage side but imho this is a regression in glances. These users aren't disconnecting sensors, they're just letting homepage ping the glances api over time, just like it has always done. But seemingly unpredictably, sometimes the api sends nothing back, and as another user reported the entire sensors section disappears temporarily in the glances dashboard too. Again, to me this seems like unexpected behavior. Ya know, imho |
|
Same issue here +1! Except for me its Dashy + Glances, with Dashy being unable to fetch the data.
|
|
Also have the issue with homepage + glances 4.x. Ended up rolling back to 3.4.0.5 for now. Thank you to everyone for the attention/logs/debugging to help get this resolved! |
|
@Matt-UKTX @FurkanVG |
Latest :dev with version 4 flag in glances yaml config did not fix anything for my side (same issues like the screenshots above) also @RazCrimson idk how i should run your commands if i run my glances via docker? |
|
Also tested the dev build via Docker on Unraid against homepage and no change. Behavior remains the same. |
|
@FurkanVG Could you try this command? The logs should automatically start showing up when you run this. |
|
The issue persists for me on |
|
@Matt-UKTX Could you try the above command and provide logs? |
Here you go... 2024-05-15 04:06:54,117 -- DEBUG -- Cannot connect to an HDDtemp server (127.0.0.1:7634 => [Errno 111] Connection refused) Now here is something weird, when I go to homepage now, the only sensor working is the temp sensor and the rest are missing. This is what I see in glances web interface: |
|
You mentioned that u had Glances had Temperature data earlier in v3. Other plugins/items in dashboard should go back to normal once you remove the below portion from the above command We had disabled other plugins to prevent unnecessary entries in the logs, that is causing the dashboard to fail. |
When I ran the |
No they are not the same. Since I can't reproduce the issue on my end, I just used I didn't want to make a release hastily like I did with I'll go ahead and make a new release if the changes in |
Makes total sense. Thanks for clarifying. It certainly appears to have fixed the issue for me. Please let me know when the updates get made to |
Just tried the newest :dev, no difference. Here are my logs from your docker run command: |
|
@FurkanVG What issue are you facing on latest dev? Homepage is not working with those If the issue is with homepage, can you just try the Note: Use my command to only generate logs. If you try to use the container spawned by that command with homepage, you will have issues with homepage as all other plugins are switched off. |
Im running :dev on glances. A snipped from my homepage services.yaml using glances: |
|
@FurkanVG |
|
@RazCrimson that works fine: |
|
So things are working fine on glances end, but homepage is still failing. 🤔 @FurkanVG |
|
@FurkanVG Found the problem. |
|
First item in the troubleshooting guide: https://gethomepage.dev/latest/more/troubleshooting/#service-widget-errors As I said, this issue is a bit confusing because its about two separate issues, but I think the actual OP here just has a config issue or something like that. If OP is having an issue setting up homepage with glances they can open a discussion / ask our bot. Again, perhaps this one should've been closed, vs renamed to be about the sensors issue. Either way, thanks Raz for all the work on the sensors issue |
|
Well the sensors plugin was indeed a regression from the glances. Anyways, @Matt-UKTX You can switch back to Closing the issue as nothing more to be done glances end. |
Funny. with Glances 3.X, it worked perfectly fine having the "/" :D I had not changed my yaml for months, and i guess V4 handles it much different than V3. Thanks @shamoon. Everything works now. |
Hi, has |
|
I’m using the latest image as well, and have tried the most recent dev image and still see the issue intermittently in homepage across my three Synology units and my Unraid box. The scenario for the repro is navigate to the tab where I have glances data exposed, wait for 30-60 seconds as it updates and then it throws the same errors noted in this bug. Not opposed to filing a new bug as this is closed. |
It appears that |
Describe the bug
Firstly, i rarely do bug reports, so spare me for mistakes.
I am using gethomepage.dev for my personal purpose. In this im using their Service Widgets of Glances (https://gethomepage.dev/latest/widgets/services/glances/) to display some stats from Glances on my homepage:
I also use Watchtower to (mostly) auto-update my docker containers. This includes glances too. Today morning, my glance docker stack got updated, which broke it:
I have some tech knowledge, but not much. The error message in Homepage is giving me some weird infos:
If i check my Browser console, i get lots of Errors. I took out one snippet:
My Homepage is trying to call this URL and it returns into an Error
https://home.XXX.de/api/services/proxy?type=glances&group=FURKANS-KALI+Performances&service=Memory+Usage&endpoint=3%2Fmem&query=%7B%22refreshInterval%22%3A1000%7D
Im guessing now, that Glances 4.X did some reconstruction of how the API works aka how some values can be called. For this to be fixed, gethomepage.dev would need to update their glances modules?
To Reproduce
Steps to reproduce the behavior:
Expected behavior
As stated above in the screenshots, my homepage should be able to fetch the values without problems.
OS:
Description: Debian GNU/Linux 12 (bookworm)
Release: 12
Codename: bookworm
Raspi 4B
Like I said, i think this is an issue on homepage side, but before i go their, i wanted to ask here if this issue is known.
Thanks!
The text was updated successfully, but these errors were encountered: