Hello All,
Not sure if this is a bug with Webmin 1.870 but this didn't happen until upgrade from 1.860.to 1.870
OS Debian Stretch (9) 64-bit
Randomly and quite often i am getting "lost network connection with (hostname):10000 " This has happened ever since the Webmin upgrade to 1.870. Also noticed that prior to this happening the CPU starts maxing out( please look at attached screenshot.) Makes no difference what module Webmin is open to. Also in screenshot you can see with top running multiple instances of '/usr/share/webmin ,,,' consuming a lot of cpu.
Trying to get as much info as possible, as soon as this happens I can in fact 'telnet localhost 10000' successfully
Have used Webmin since the first version so very familar with the working of Webmin,& it is one great Admin tool,so thanks to all who does input on this !
This is the first time this has ever happened with many many versions of Webmin installs on mainly Linux variant machines.
Thank You,
This is the theme bug.
Duplicate of https://sourceforge.net/p/webadmin/bugs/5059/
It's been fixed few days ago. Just ago and Force Update the theme from its settings to latest developement version.
I'm very sorry about that.
Illa,
Thank You very much. Tried seraching before posting. Guess I didn't search enough,
Thank You for a definitive fix for this scenario.
Barry
Not sure if this will help you any. I tracked down on my system what when the crash was happening and what was causing it. I had upgraded to ubuntu 18.04. It apparently disable SSL and the default-ssl apache server was also disable. When trying to go to a url that had "https" it would cause webmin to go off line. I know webmin doesn't run on apache.
Either way I was able to fix my issue and it isn't happening now. I seem to remember on my son's server he had a similar issue when his SSL expired.
Hopefully that helps you narrow down what's actually causing it. So you know I'm on version 1.890
My webmin has become incridibly slow (somewhere between 15 seconds to a minute for pages to load) and I'm getting the "connection lost " error a lot since I've upgraded webmin from 1.890 to 1.900 yesterday. 1.890 was working fine. The server is Debian 9. Apache (2.4.25) is running and ssl is enabled (mentioning it because it was mentioned in the previous post). I was using the "Authentic Theme" version 19.20, after reading bug report I upgraded to 19.22 and when that still didn't improve things I upgraded to 19.23-beta3, which seems quicker, but still slower then webmin version 1.890 and I still get "connection lost" errors, although less frequent it seems. Console shows a lot of stuff "pending" for a long time and sometimes failing. The "Old Webmin Theme" does the same, so I don't really think it has to do with the theme.
You could disable real-time statistics in theme configuration for slow connections. You should also not disable caching.
Try restarting Webmin and closing/reopening your browser.
I'm on (at least) a gigbit network here, very fast and very stable. Disabling real-time statistics does nothing, pages still load slow or get "lost connection" errors. Caching is not disabled. I've tried different browsers and even OS's. Tried Firefox and Vivaldi on linux and Firefox and Edge on Windows.
I'm seeing a lot of css errors on the firefox and vivaldi consoles about "bundle.min".
30 seconds for initial load is bizarre. My initial load (doing Ctrl+R) for my server (bare metal, not VM) takes 999ms. For VM on my Desktop it's only 5 seconds for when loaded as a bundle, and 3.5 when loaded each script separately.
I think, you should try disabling AddOns (Like add block and some other), or try Incognito Mode (Private Tab).
Those CSS errors are harmless.
The reason why it's too slow, regardless of the theme (like you said above), is too little resources given to Webmin. Are you running it in VM? Anyway, what are the specs of the server, where you're running Webmin?
How many resources does webmin require? What are the recommended requirements?
It's a test server on which I tried the upgrade before doing it on the production servers, thankfully, else I would now have a unworkable production environment.
The server is only running squid and apache (barely). There are 2 client test machines connected, so the load is almost non existent. To make sure those are not in the way I've stopped apache and squid, so now webmin is the only thing running on this server. The host is vmware esx and specs are 2 CPU cores, 2GB memory (~1,5GB free), 20GB disk (~11GB free)... with webmin running and squid + apache disabled. Still there is no change in the functioning of webmin. Still slow, still getting "connection lost" errors.
I just tried purging webmin from the system and installed version 1.890 again. That is now working slowly as well, so I upgraded it to 1.900 again, which also remains slow.
What else can I try? What information can I provide that might help? I'm willing to try anything you throw at me, but I'm no web developer, I need help. I'm not going to upgrade my production servers before I've resolved these issues on the testing server.
At first glance, what looks strange to me is that you have 96% of CPU load?
Second, I would check networks configurations. If you connect to test server using proxy, try direct connection. Check your DNS and ping to other servers. Check the ping to the test sever that is running Webmin.
Try disabling/checking Firewall.
Try purging
/etc/webmin
with fresh install to see if anything would change.As last idea that I have a the moment, try running
update-from-repos.sh
from/usr/share/webmin
to update to the very last Git version of Webmin, and running post install scripts.Indeed, there might be tons of reasons why this could be happening, and this is why it's slow and unproductive way of doing this via the ticket, without actully having an access to the server.
It's not Webmin issue apparetnly. We have no such issues on many different machines, in many different configurations.
I think you're confusing the problem of the original poster with mine. I don't have 96% CPU load... it doesn't come above 20% on my system. I've atteched a screenshot of the dashboard. I'm also not connecting through a proxy, the server is a proxy itself. I connect to webmin directly. I can ping google.com with an avarage time of 3ms and all connected hosts I can ping with times of around 0.3ms, so I think network and dns are fine. Sites served through apache are not slow. The login page of webmin also appears very fast. Only when I login to webmin, the webmin pages become slow. So I think it is a webmin issue.
Also the 2 production servers (which still run webmin 1.890) are built exactly the same way (running apache and squid too) and they don't have the problem. That is why the difference is so very obvious to me... the old webmin works like a charm on the production servers, but the new webmin does not work very smoothlyat all on the testing server. I've included a screenshot of the webmin dashboard.
I again removed webmin from the system, making very sure to remove everything related to webmin. I installed version 1.900 now and upgraded the Authentic theme to version 19.22. Have not seens a 'connection lost' error yet and most pages load fairly quick, but some of them still take more then 30 seconds.
Chris, what about
/var/webmin
directory? Did you try purging it? What are the log sizes? What is the size ofmodules
directory? Are there any particular errors found in logs (miniserv.log/miniserv.error) when going to the modules which load slowly? In case the directory size is to large, save it first.What modules in particular do you experience slow load issue? What modules load fast (normal)?
Besdies all, to be sure that your issue hasn't been addressed already, please try installing latest development version of Webmin. I would simply tried installing latest version from Git by running
update-from-repos.sh
script.Chris, for your information, you can significantly enhance initial load speed by enabling compression in Webmin/Webmin Configuration/Web Server Options and choose Use pre-compressed file and compress dynamically for Gzip compress static files?.
Chris, we have done some optimisations.
Please install latest Webmin from Git using update script, and latest theme (update script can do it), and see if it works faster for you!? :)
Sorry for the late reply, I did not really get around to trying anything yesterday.
Increasing the buffer size seems to have worked. Pages load quickly again and I don't get 'connection lost' errors any more.
Today I've done a clean install of v1.900 and updated that to the latest version with the update script, the authentic-theme too.
First I did the regular 'apt-get purge webmin', then I did a find on the entire disk searching for webmin and removed any lingering files; that also removed /var/webmin.
The sizes after install are very small:
58K miniserv.error
26K miniserv.log
1.4M /var/webmin/modules/
The logs show no errors that say much to me, but find the logs attached.
Before, most pages were slow loading at first, but some were quicker when I revisited them. After I waited a few minutes, they were slow again. Things like the file manager and 'Partitions on Local Disks' took over a minute to load. I also enabled 'pre-compressed file and compress dynamically' in the old (slow) environment, but that didn't change things.
Now pages just takes a few seconds to load. I left 'pre-compressed file and compress dynamically' enabled in the new enviroment as well.
Thank you very much for the great support, you really helped me out a lot!
Keep up the great work! Webmin really makes linux systems useable for non linux users, which is our main use for it here. I love how easy it is to give users exactly the correct permissions.
Thank you for your kind words.
We're very glad it worked for you!
Hi Ilia, I've noticed the webmin interface became slow again when I upgraded the the Authentic theme from 19.22 to 19.31. When I select the Old webim theme it works properly again, so that would suggest something with the theme. The buffer size increase from the above patch is still in place. The problem seems a bit like the one here: https://github.com/authentic-theme/authentic-theme/issues/1287
Now I don't run a machine with kvm directly, but it is a vmware esx host on which I installed debian. I've attached screenshots of the waterfall.
Check the options for Web Server Options under Webmin/Webmin Configuration.
In particular, check Gzip compress static files? option.
What is it set now to? Try setting it to different options and reloading the page.
Are you sure that you really applied the patch and restarted Webmin? Doublec check, can you find in
miniserv.pl
the string 1024?Another test case, is to try to download directly from your server, how fast is it going?
Gzip compress static file was set to Use pre-compressed file and compress dynamically. Changing it to eithe Only if pre-compressed .gz file exists or Never does not seem to change much. I've double checked miniserv.pl for the following patch:
https://github.com/webmin/webmin/commit/808ac0cc75b7972193d70be1fdea46c52b7b3829
It's still there. What do you mean with string 1024? the following block can be found in my miniserv.pl at line 1024:
I just meant that you need to make sure that you are really replace 1024 to 32768.
Did you try to download/upload directly from your server, how fast is it going? Try using Upload and Download module.