Open adamgins opened 9 years ago
I have this issue too, any ideas?
I am about to try and bypass the issue with Cloudflare. I'll let you know if it works.
@bitomule similar issue with Cloudflare... it just took longer to get the broken link
I'm not even using nginx. Just Cluster and trying to connect to the server throws that error only on chrome, safari just shows the broken image.
@arunoda any thoughts, pls?
I didn't got these kind of issues. May be since I don't use longer files. Could you please create a isolated case for me to work on?
@arunoda thanks. Will try and setup something... not that simple as it's tied to s3. @bitomule were you experiencing the issue with AWS s3 or using another CFS connector?
@adamgins yeah, I'm also using S3.
@arunoda trying to setup a test environment for you... what's the best way to do this? I have a base project that I have deployed to meteor.com but not sure how to setup cluster in that environment currently it's running at http://tests3cfs.meteor.com/ (No cluster)
You can just drop an image on and then it shows it below... so pretty basic.
I have the code here https://github.com/Buzzy-Buzz/meteor-debug note I have taken out the s3 keys in the Images(collection/images.js) . It's just a test bucket, so I can send you this privately if you email me adam.ginsburg@buzzy.buzz or you can use your own if you have them.
Also, as I understand it, I cant get this running with the graphicsMagick transformations on meteor.com (it has to be installed on the server). It's these images that seem to mess up. @bitomule not sure if you're seeing this too? ie is it with the transformed images only or the original ones?
thanks for the help
I think there's a big issue with cluster and collectionFS, I can upload or download files and the same code works outside cluster. @arunoda is there anything in cluster that can be blocking access to S3?
In my case I solved the issue by increasing the send_timeout
in Nginx.
Hi, I am using Cluster with nginx reverse proxy (Docker install thanks to Sunsama). I am using nginx to serv files via s3 and I am getting some random errors on serving files ons3 via collectionsFS. I am just struggling to debug/pinpoint where the actual issue is. As Cluster is part of the equation wondering if anytone has any thoughts on this, please?
Basically, sometimes, an uploaded file will initially show a
200 OK
and then a few seconds later show anet::ERR_CONTENT_LENGTH_MISMATCH
in the browser console.So the image shows up and then disappears a few seconds later.
on the server side nginx is throwing a
[error] 26#0: *759 upstream prematurely closed connection while reading upstream,
I think this is just nginx saying there's an issue with the backend and the error is deeper within the server.
Is there something in Cluster that would be doing two request, where the first is OK and the second has an issue?
Here's the detail of the error logged with CollectionFS which has some screenshots and additional info https://github.com/CollectionFS/Meteor-CollectionFS/issues/690