I’ve been trying to track down this problem for months. I can’t seem to get our desktop clients to connect reliably with our server and I can’t seem to figure out why. Our trouble shooting has lead us to install a new internet provider (dumping our cable connection). We were reluctant to replace our internet connection seeing that we were having no problems reaching any other service co-located with our nextcloud server. The replacement of service certainly helped the problem (we now have a couple of machines that connect reliably to our server throughout the day) but other systems in the same office will not connect. The reason that we identified our cable internet connection as a problem was that when no systems could connect to our NC server from our office we could always connect through a neighboring business wifi connection. Despite the fact that the problem markedly diminished with our new connection we are still having problems. We have tried a number of fixes:
Rebuilt our server and replaced the hardware (Odroid)
Replaced our cable internet connection, router and cables in the office (the remote site)
Our NC server runs on an Odroid-c, Ubuntu 16.04, Ngnix and rarely show any load beyone .03. Our server is connected to the internet via a comcast cable connection. We run a mailserver and webserver on this connection and never have a problem reaching those servers. Seeing that replacing my cable connection at the remote site partially solved the problem I’m wondering if I should replace my connection at the server site…
The link below nginx_nextcloud error log… there are many timeout errors… perhaps someone can take a look and see if they see if something is out of order.
These timeout errors should be handled. Review your fastcgi-timeout settings and if you think they are already long enough, you have to look into more details which processes take so long.
If you haven’t set up any caching mechanism yet, do this first, this reduces the load on your database and the waiting time for it. This is important if you have many people using it. You are already using system cronjobs?
Thanks for the note . There are only 5 users on our system. The system load is always sub 1. I’ve looked to see where I might lengthen my fastcgi timeout but I don’t see a variable for timeout… I looked in fastcgi.conf and fastcgi_params.
I’m on a computer in my office (direct wire to nextcloud) but my desktop client will not connect. The system load is .11 and there are no other users connected to nextcloud. However, webdav will connect fine. This is maddening.
What is a bit strange is the 499 error which means that the client gives up before the server can send data. So it could be client-side as well (network connection, firewall, …), you could try to connect one of these clients to one of the test servers: Demo instances of 9-14 now live