Uploads to NC 15 (latest) instance of Official VM image that have many files (see examples below) consistently fail. Usually no error from UI, nor in any logfile I can find (high verbosity in all logs). VM installed basically with defaults no extras, powerful server not doing anything else. I can successfully upload these test files to a commercial NC server (version unknown) on owndrive.com
No issues uploading single large files or other uploads that don’t have especially large number of individual files, all uploads tested over https.
So there’s no error usually, usually just the upload progress meter is stuck midway…once in a while I will get an “An unknown error has occurred” shown in the UI.
I can watch the apache access and redis logs things seem to be humming along requests are being processed then suddenly everything stops, it almost looks like redis stops a second before apache does…
There were a few things the redis log was warning about I tried changing such as increasing
echo never > /sys/kernel/mm/transparent_hugepage/enabled
Tried increasing redis keep alives, and a few other redis parameters with no change.
php-fpm I upped these to quite high, no effect.
In NCs config.php I tried changing
from redis to use APCu, no effect.
Also upped the ‘memcache.locking’ => ‘\OC\Memcache\Redis’, timeout from .5 to ‘timeout’ => 1.5 no effect.
I’ve stopped at this point because I’m just shooting in the dark now.
Ubuntu 18.04 3ghz 8 core server 16 gigs mem, Raid 10 - doing nothing
VM imported to virtualbox with defaults, running headless, private network self signed https
Uploads, using either of extracted below to test:
Testing with both latest firefox and chrome, no extensions installed.
NC Logging in web admin
Apache access & error
PHP (had to enable)
Redis (increased verbosity)
Postgres (enabled verbosity increased)
Thanks for any help!