[Resolved] Service Incident Update: Projects Loading Slowly or Not At All, NPM Reinstalls, & Missing Code

Having the same issue and I’ve tried every solution in this thread, but none of them works.
Here’s also me doing df -h / /tmp /app
image

1 Like

Wow, 100K free is not gonna be enough to install dependencies. Thanks for the screenshot.

1 Like

Screen Shot 2020-06-21 at 1.39.23 PM
@wh0 is this enough?? lol

yeah. might be worth trying npm instead of pnpm, with all that space you @Daniel have in /app

OMFG I TYPED enable-npm AND NOW I CANT REWIND AND MY SOME CODE GOT REPLACED WITH BLACK DOTS!!! @wh0 DO YOU KNOW HOW TO FIX THIS!!!

oh geez, this?

Yes it got replaced with those things. PLEASE HELP!! IM ******* PANICKING

and you said earlier that you deleted the .git dir? do you have any backups?

I have one back up on my computer from June 14

Ok I just checked and it’s only some files so hopefully I’m not totally ******.

4 files are just replaced with dots :cry: good thing I backed up my project. I think my .sqlite files are fine and I should be able to paste in the code of those 4 files since I haven’t changed them since June 14.

1 Like

Code pasted! Everything is working now!!

@wh0 im getting this > UnhandledPromiseRejectionWarning: SqliteError: disk I/O error
now. Do you know to fix it??

I’ve been seeing this intermittently. That is to say, after many subsequent attempts, a later disk write would succeed.

image

Above: a cluster of I/O errors observed last night. red-orange = I/O error, light blue = success. time on vertical axis, more recent entries on top, ~1 second per pixel. longer bars mean longer write latency.

But I wasn’t using sqlite in that case. I don’t know if the error would corrupt the database. (In my case, I was rewriting the entire file.) It could even be that these I/O errors are what cause files to become replaced with dots, e.g. when the Glitch editor encounters such an error.

So at some level, maybe the solution is to do your best to back up the database and try to rebuild your app to store its state on a third party service.

Or if you want to try something wild, just retry the write until it succeeds. Again, I don’t know if an error would corrupt the database.

I have the same issue :

node v12.0.0, with pnpm

9:01 PM

Installing…

9:01 PM

Performing headless installation

9:01 PM

ERROR ENOSPC: no space left on device, open ‘/rbd/pnpm-volume/store/2/store.json.2640651246’

1 Like

@wh0 should I just try uploading the old sqlite file from my backup?? Don’t really care about the data since it isn’t important anyway. Do you think it’ll work?

Welp, I’ve got the exact same issue with pnpm…

could it be cus of the outage?!?!?

It no doubt is caused by the outage, so we’ll have to wait for Glitch to fix this…

Wow, that’s a good situation to be in. It’ll save you the trouble of figuring out the details of how sqlite works. If you can upload the backup version of the database–and it may take multiple tries if there are intermittent errors–then you can be sure (more sure anyway :roll_eyes: so much uncertainty right now) that whatever’s saved is not corrupted. At least not corrupted until the next write attempt.