I’m seeing an issue where large uploads appear to fail (OOD 3.0.1). I’ve been testing with a 3GB file and the file upload progress gets to 100%, waits for a while, and then I get ‘upload failed’. However the file appears to actually have been uploaded and is complete (compared checksums of original and uploaded).
Although it was not apparently the problem, I was seeing log messages in httpd log like below. I resolved those by adding a ProxyTimeout 3600 to our httpd conf.d/ood-portal.conf but still the file upload error persists. I’m mentioning it just for reference, as resolving this issue did not resolve the file upload error.
“(70007)The timeout specified has expired: [client xxxx:42816] AH01102: error reading status line from remote server httpd-UDS:0”
I’m not seeing any indication of error in the ondemand-nginx logs either. This appears very similiar to an older issue File Upload Issue but that was supposedly fixed in a much older version of OOD. I also checked the settings for max file upload size and we haven’t altered the default (around 10GB). Any ideas where else to troubleshoot this issue?
The upload widget in browser is giving an error of ‘upload stalled for 128 seconds, aborting’ when this fails. I didn’t see it before because the error message quickly disappears but I was able to get a glance at it finally. This seems related to the uppy timeout issue mentioned in PR extend the timeout for uppy uploads by johrstrom · Pull Request #1600 · OSC/ondemand · GitHub since the fix there is setting it to 128 * 1000 (128 seconds). Maybe a longer timeout is needed? I tried changing the uppy timeout in /var/www/ood/apps/sys/dashboard/app/javascript/packs/files/uppy_ops.js but it doesn’t seem to be reflected in the browser…is this the right location? Do I need to do something to get the updated javascript to be used?
Have you resolved this issue now? I noticed that the default setting in version 3.0.1 is 128*1000. When I upload files larger than 3GB, it shows a failure message, but the upload actually succeeds. Initially, I couldn’t see the file, but later it appeared automatically and was intact.
We have not resolved the issue. I notice the same behavior as you with large files where it shows a failure message but the file does succeed in uploading and checksums correctly. My last work on the issue was an attempt to change the setting 128*1000 to be larger in uppy_ops.js but in order to do so I believe it’s necessary to recompile the js (bin/recompile_js in the dist) and I had issues with getting ruby to find my embedded js compiler…I haven’t gotten past that point.