Client synchronization issues

Desktop
1

I was having a difficult problem using my owncloud.

description:
Company colleagues like to use their owncloud client [not WebDAV] to synchronize their owncloud server resources, they think this feature is great and very convenient. However, when a colleague or manager to the company to share more than 500M full file, nearly 100 owncloud client at the same time pull the shared resources of owncloud server to everyone's computer, this time, the server pressure surge, memory error : Memory overflow; disk io stuck, resulting in the server crashes. Of course, when this happens, I installed the server version of owncloud is: owncloud9.0.1.

The server model number is Dell 2950
CPU: Intel quad-core (specific models do not remember)
Memory: 16GB
Hard Drive: Five 1TB SATA drives (configured with RIAD5)
Operating System: CentOS 6.6

Now I re-upgrade their owncloud9.1.1, and now we do not share the full capacity of large files, so the normal state.

I would like to ask, there is no official solution for the problem of change?
For example: system optimization? If you can solve the problem, how to optimize the operation?
Or is there any other solution? For example: load balancing and so on.

thank you very much!

2

The official docs say that environment like yours are up to 150 users (it certainly depends how it is used): https://doc.owncloud.org/server/9.1/admin_manual/installation/deployment_recommendations.html

As you can see, with load balancing the setup gets much more complicated and you need more hardware. Did you do some optimization? Redis-cache, mysql-cache, webserver, ...? What process is responsible for your memory error? Perhaps you need just a bit more memory. Perhaps also worth investigate the disk-io-issues (not sure if this is a normal and expected behavior). Running the system and some caches on SSD might help. You can also check some general guides for running high-load-systems on apache. Sorry for this bulk of more generic tips, perhaps someone in a similar environment can give your more specific tips.

3

What is the memory usage of the Apache processes during this?