I set up Nextcloud AIO on a 2 GB Linode instance, and I notice that the instance consistently runs with all the physical memory and swap saturated to nearly 100%. See picture below from the Nextcloud > Settings > System view. The Nextcloud instance works well in general, but every 30 minutes or so, it just becomes completely unresponsive for a few minutes before returning to normal again. I am wondering if this is related to the high memory usage.
Consider adding the point of user to system resource ratio too.
Higher then number of simultaneous users and installed apps, more system resources you may need.
For example, my nextcloud installation runs fine with 2GB RAM and 512MB SWAP. Memory usage is around 1GB with no swapping in idle times, but when there is load of many concurrent users, system usage shoots up dramatically !!
At peak time, I have to allocate upto 16GB RAM to sustain the load for a responsive performance !!
It’s the same on my side. I’m also running Nextcloud AIO in beta channel (Master-Container version 4.5). It’s sitting inside a Proxmox VM on an Ubuntu 22.04 LTS, which is up-to-date. I installed the QEMU guest Tools and had it running with 16 GB of ram, just like you did.
On the Proxmox interface it says that it’s consuming almost 15 GB ram, but when I run htop in the container it’s a lot less - hardly 3 GB.
I also tried to switch off the ballooning function and set it to 8 GB. The behavior was the same. In the Proxmox UI it’s almost at 90% Ram usage, and inside the container it’s hardly 50% of it.
As you see this is the graph showing 16GB at the beginning and 8 GB RAM late on:
I’m also seeing this issue of very high memory use in Nextcloud AIO consistently when uploading mp4 video files from an Android phone using Auto upload to a folder to External storage on SMB/CIFS network share.
When uploading gigabytes of mp4 files, the VM’s memory fills up after every few gigabytes uploaded eg. in my case after about 5GB of videos uploaded 16GB VM memory easily fills up completely causing the VM to stutter and eventually become unresponsive.
I’m running Nextcloud AIO on an Open Media Vault 6.3.7-1 (Debian) VM instance under Proxmox pve-manager/7.4-3/9002ab8a (running kernel: 5.15.102-1-pve).
I want to say, if you have
ElasticSearch, Collabora, then 2GB memory might be too less.
ClamAV is recommended 4GB memory, absolutely, I disabled it.
But full text search and office at least require 2GB or more to work well.
Actually, if you have many photos and enabled imaginary, at least another 1 or 2 GB memory is required.
Another experience I have is HEIC bug in imaginary and cost much memory. But this is fixed or improved after NC26 or NC 25.0.5.
Just for your reference.
My home setup use similar AIO and given 8GB memory as I hosted >100 GB photos.
I’m also trying to find the relevant logs that shows the file uploads.
Can you please give me some pointers where should I look for those logs? Which container handles the actual file uploads? I don’t seem to find anything related in the container logs I’ve checked eg. docker logs nextcloud-aio-nextcloud.
I’m on Nextcloud 28 (Turnkey Linux Appliance; fully updated as per today).
Back from my vacation, I have been trying to upload several GBs worth of fotos and videos. But an out of memory error consistently has been killing off my apache2 process. I have increased the VM’s ram from originally 2GB to 4GB, to 6GB and now to 8GB. It’s no good.
When looking at the status of the process, systemctl tells me that various clamscan processes remain running. So my guess is that clamscan is the culprit.
What should I do? How much RAM is needed? This is a small personal NC instance with only three to four users but with lots of GBs worth of data.
