For the 60s: it seems to be a default proxy_pass setting in nginx:
http://nginx.org/en/docs/http/ngx_http_proxy_module.html#proxy_read_timeout
And in older post, there were already problems regarding the avatars:
For the 60s: it seems to be a default proxy_pass setting in nginx:
http://nginx.org/en/docs/http/ngx_http_proxy_module.html#proxy_read_timeout
And in older post, there were already problems regarding the avatars:
Check out mtr
(my traceroute). That tool shows me 85% packet loss from 4.68.38.173 (the 4th hop), and 0% packet loss from all others. I like to imagine that router is somewhere deep in the Atlantic ocean.
Iām in Australia and help.nextcloud.com has been slow as can be for years, Iāve more or less given up on hoping it would ever be better. The only reason Iām here, is today itās slow beyond. My browser has been sitting on:
for literally hours. And trying it through a US proxy yields same result. The only way I get to post this at all, is on the TOR network (using the Tor Browser) where itās merely very slow, but functional.
Nextcloud rocks, it really does, but how can it be the only service among the hundreds we all use that has a help site that is so non-performant? Itās quite hard to believe really.
Some improvement have been performed shortly. Please double check if logout and clear browser local data improves forum performance (try when not logged in/using private mode as well).
For me it did the trick - loading posts in new tab in firefox dropped from 5 sec to 1 secā¦
What was changed? I did a quick cold-cache test (used a private window) and it seemed about the same (slow).
I know about some DB optimization - but this should have no affect on web interface sideā¦
it looks MTU is even lower than reported before
on Windows 10
ping help.nextcloud.com -f -l 1172 -n 1
works
ping help.nextcloud.com -f -l 1173 -n 1
fails
but IMHO it has no effect on āgoodā connectionsā¦ I have VDSL2 account in CH with 100M+ with MTU 14xx+ā¦ and see almost no issues nowā¦ Iā not sure this is the reason for reported issuesā¦
for me forum works faster since some days/weeks
UPDATE: the low MTU was a config error on my side (MTU 1200 remained active on my client after some troubleshooting). Effective MTU is always 28Bytes lower than MTU of my client - but I see no difference when I configure MTU 1200 and MTU 1480 - both values are fast enough now.
I was to post about this issue in a new thread but only to procrastinate post it here, yes, this website is hyper slow. It takes up to 5 minutes to load, and sometimes I have seen it takes more than 15 minutes to load any page. I have to use the Google cache to see what I need from this website.
If you still have problems, @daphne reaches out for users that still have issue. They already fixed issues that helped some people but they need feedback from users that still have issues.
@neoacevedo
Can you use the browser dev tools (F12) and there Network and post a screenshot. You can find details at entry 64 and an example at 65 of this thread
Something else to try: delete all browser cookies, cache, and local storage for help.nextcloud.com
. I was experiencing slow loading in Firefox just now and it loaded very quickly after I did this.
this is exactly what I found useful a month ago - does it mean today you see reasonable load time? what are your current numbers?
Ah right, thanks for the reminder!
And a huge thanks to you and @tflidd and @daphne for helping track this stuff down.
It means the forum is slow yet usable. That is, 5 to 10sec until the āloadā event fires.
Your suggestion of clearing site data seems to relieve a problem that (intermittently) makes it take more than 5 minutes to load or timeout and sometimes never finishes loading.
I still have the other problem where download speeds for Javascript assets are slow over my home 1gpbs fiber connectionāabout 10KiB/sec. Some other folks have reproduced this but itās an odd one. Does not repro over a mobile data connection. Luckily this slow download speed is only a problem when I try to use the forums with a completely cold cache, so, not too often. With a cold cache Iām waiting a few minutes until the āloadā event fires.
Hereās another example of the slow download speeds:
$ curl --compressed -o /dev/null https://help.nextcloud.com/assets/discourse-12a807403b8f23c14a92099c4e078ddbbac432d1409edb731c4150ac17a7d55d.js
% Total % Received % Xferd Average Speed Time Time Time Current
Dload Upload Total Spent Left Speed
45 490k 45 224k 0 0 9163 0 0:00:54 0:00:25 0:00:29 9521
great you can use the forum now!
but the issue loading .js make me completely curiousā¦ my load times are sub-second on native Win 10 and even Debian VM running in hyper-v on my laptopā¦
sounds really weirdā¦ while this is a proof the forum āitselfā works OK and the problem must be related to the provider: router/providerā¦ I have no good idea how to troubleshoot further.
BTW: Few days ago I found something was wrong with Firefox - private window with ādisable cacheā check-mark tick never loadedā¦ MS Edge worked fineā¦ - today itās āback to normalā - FF 107 loads the forum in a private window with cache disabled in ~3sec
Hello,
We changed some settings with DNS and some other stuff with the Forum engine itself to keep it more performatic. We ask now to wait up to 24 hours, than clear DNS cache and try to see if the problem is gone.
Note: This is happening only with few users, we did some tests and depending on the ISP it gets slow (the very first access only) than it becomes fast.
Letās wait a bit more, and we hope the issue has gone. The DNS change was done today 30-11 at 2PM CET.
Hello All,
After some settings changed, we noticed that for some users the slowness in the very first access has gone. I kindly ask you all to test it now again to see if the issue has been solved on your end also.
As this problem happens only with few users, it should be something on your end.
Ok, If not solved yet, I would recommend to check and change the MTU size on your router based in the test we will guide you to check the ideal MTU size in your case and prevent data fragmentation:
If you use Linux, the way to do that is using the ping command:
Start with this one:
ping -M do -s 1492 help.nextcloud.com
Check the output, if it says that youāre having fragmented data, slow down the number (1492) in 10/10 for example, next ping should be 1482, like this:
ping -M do -s 1482 help.nextcloud.com
Do it until you can complete the ping, most of cases the number should reach 1462 (or near) to you be able to ping the address succesfully without data fragmentation.
Reaching the exactly number, add +28 on it, for example, if the number that pings the address was 1462, add +28 and youāll have 1490. That should be the MTU size youāll need to adjust on your Internet Router (each one has different places to find this info, search on the web where to find yours if itās a bit hidden from you).
Change the MTU number to one you find, restart your router and test again.
PS: If youāre using Windows, the ping command should be something like:
ping help.nextcloud.com -f -l 1492
And slows down from this value till find the one completes your ping request.
Donāt forget to add +28 in the end to setup correctly the MTU size on your router.
ā
Itāll solve not only the access to our help.nextcloud.com forum, but in all other sites having the same issue over the internet.
Let me know the results and your thoughts on that,
Kind Regards,
Thanks for the MTU tip! This is super fascinating.
I just tried the tests Iāve been doing, again. Forum performance is pretty much the same as Iāve described in earlier messages above (and in other threads). That is: (1) sometimes it wonāt load even after waiting for several minutes and (2), slow transfer speeds (~10kbps), reproducible with, for example:
curl --compressed -o /dev/null https://help.nextcloud.com/assets/discourse-90317d3fc0cd0d71125e07cdb6bf5b38f87dcbdf5e380a1e5e4734435256feb9.js
Clearing all cookies and site data fixes problem (1), and using mobile data instead of my home network fixes problem (2).
Can you say a bit more about your understanding of the problem and how MTU might be the root cause? I asked a couple networking experts and they were very skeptical that tweaking MTU values would help.
Iāve never changed any MTU settings and I canāt think of any other websites where I see slow transfer speeds like I do with help.nextcloud.com
, so Iām both skeptical and excited to learn from this.
I found a setting for MSS clamping in my router config. I canāt find a place to change MTU. I have a unifi udm pro. Is MSS clamping the same as setting MTU size?
This site tells me I have MTU=1492, MSS=1452, RWIN=64256:
https://www.speedguide.net/analyzer.php
Also: what were the DNS changes exactly? I donāt recall having any DNS problems on my end, for what itās worth.
Also: my home ISP connection uses PPPoE. Not sure if that makes a difference.
Still seeing a very slow load on first access after clearing site data, and unending loading after a few close and restarts of the browser. Can the site admins please check that the css is all sourced via the same CDN as the rest of the site?
I reduced the MTU on my router down to 1492 (it was at 1500), and I get the same results, nearly 2 minutes to load the front page for the first time:
73 requests
1.87 MB / 846.43 kB transferred
Finish: 2.05 min
Reloading the page is much faster, without the minute and a half spent watching the coloured balls:
50 requests
350.03 kB / 10.41 kB transferred
Finish: 32.79 s
I wasnāt seeing fragmented packets in the testing anyway, so that likely wasnāt the issue.
hopefully this is sorted now: