Well, you have now shouted it out for long and loud enough and it obviously has been heard. Now itās time for the site admins to look into it. It is just fair to give it some time, is it?
Iām coming from New Zealand.
Here is ping to the forum.
PING help.nextcloud.com (95.217.53.146) 56(84) bytes of data.
64 bytes from help.nextcloud.com (95.217.53.146): icmp_seq=1 ttl=51 time=328 ms
64 bytes from help.nextcloud.com (95.217.53.146): icmp_seq=2 ttl=51 time=328 ms
64 bytes from help.nextcloud.com (95.217.53.146): icmp_seq=3 ttl=51 time=328 ms
64 bytes from help.nextcloud.com (95.217.53.146): icmp_seq=4 ttl=51 time=329 ms
I donāt think itās that slowā¦a little. But I wonder if there are scripts of something that are slow. Iāll try on a pc with chrome developer tools running t I can see all the pages and assets loading
What do you mean by that ???
The forum is maintained by volunteers. Be patient.
I think it is a little hard for people to follow the order of the thread here as replies are displayed as general comments. So I think some mistake a reply for a comment on something totally different.
That a VPN is changing the performance, could indicate a problem with some peering points. At the moment of the problem, some admin of the forum server would need to check the route to your IP. In a second step, they need to reach out to the provider, if they can help to improve the routing.
We should separate this issue from performance issues that could be solved by the server admins themselves.
The server is provided and managed by Nextcloud.
This site is too slow. Just Can not authorized with google account.
Since one year sometimes it is very slow for me too. I have been accessing from Spain.
For some reason. First time ever Iām getting good response from the site. Youāll see from previously replies I was complaining of almost 2 minute delays before site rendered.
But today itās been well under a minute. I donāt know what changed, but I much prefer it.
Same here, problem seems solved.
No improvement at all.
Yesterday, from Spain very slow
I ran another test via webpagetest.org (anyone can do this, by the way ā itās a free, public, and anonymous service). The test was run from the east coast of the USA. See https://www.webpagetest.org/result/211113_BiDcF1_bbeb8970aa292ef383ccb7338980cc2d/ . There was just one odd static asset (a PNG image) that took about a minute to download. It wasnāt blocking. Otherwise, the overall wait was only about 5 seconds! Way better.
Anecdotally, warm-cache page loads do feel way faster to me on desktop and mobile now. Very near the 5 seconds mentioned above. Cold-cache performance seems just as bad as when I originally ran speed tests ( see Help.nextcloud.com slow page load times ), though. It seems to be blocking for over a minute on several javascript files and static assets (png images). I also noticed piwik.js failed to load from stats.nextcloud.comāthat may be because I use an ad blocker (Pi-hole).
Anyway, this is great progress! Thank you @just and team!
I am from India, using Fiber to home and I too face this extremely slow loading issue of this forum pages.
And back to slow for me
Hi all,
My Experience:
Day 1:
Loading the forum the first time takes ages < 30 seconds and i refresh until it works up to 2 minuits.
After the first load it is fast.
Not changing anything just log out of computer.
Day 2:
Loading the forum the first time takes ages < 30 seconds and i refresh until it works up to 2 minuits.
After the first load it is fast.
Not changing anything just log out of computer.
Day x:
Beheaviour does not change.
I got used to it I think its like that for more then 2 years.
It is just a very slow server, I really think they do not want the community to use it
For me, TTFB (time for first byte) = 50 seconds for a simple 1.4kb PNG (not even a script)!
Itās really just a matter of putting a load balancer or even just scaling vertically the instance (more bandwidth, more CPU). It might be also as stupā¦ woups simple as bumping up or configuring PHP pm (static). I think itās a really bad rep for people crossing this forum when taking a decision to use nextcloud or not.
So the issue with the forums is kinda weird, the IP stack on the server is seriously broken around MTU/MSS handling. If you do not have a 1500 MTU capable internet connection (aka youāre on VDSL, ADSL, anything with PPPOE, etc), then youāll have issues with the nextcloud forums.
Usually youāll see this if your router doesnāt have clamp-mss (also know as adjust-mss) configured, and the remote site has incorrectly had ICMP completely firewalled. However in this case, the forums are also ignoring the requested MSS in the SYN packetā¦
Looking at a tcpdump here, Iām watching my workstation send a SYN packet with an mss of 1440. And then five packets later, the server sends back a 1500 byte packet! (an MSS of 1440 should have resulted in maximum packet sizes of 1480) My upstream router (which has a 1500 MTU and has the pppoe session from my VDSL landing on it) then sends back an ICMP Frag Needed packet (mtu 1480), the forums send back a few more 1500ās, and then finally start sending 1480 packetsā¦ Which is how things are supposed to work. Except then on the next fetch (i.e. each of the css files, jpegs, etc etc) we go through the whole rigmarole AGAIN.
This thing of it happening AGAIN should not happen - the kernelās IP stack should be caching the MTU to the destination IP and just using it - and of course, it should be honouring the 1440 mss (1480 MTU) in the first place.
Not sure if whoever looks after the server hosting the forums is around, but if you can check that a) youāre not blocking ICMP frag needed packets, and
b) thereās no weird firewall rules set to adjust the mss of inbound packets anywhere in front of the server?
It is not slow for everybody and not all the time. So there might be different issues and people are affected differently. There were some general problems and it was slow for everybody (many 5xx errors and some images/resources not loading and running into timeouts). Some of the things were fixed. But not for all.
I can ping help.nextcloud.com and set a MTU of 1470:
ping help.nextcloud.com -f -l 1470
I get packages of the same size back.
Oh good point, so yeah, they are actively ignoring mss in the syn packet, and the icmp frag needed packets