That doesn't really make much sense and I suspect there's some important info missing. What -exactly- are you trying to change? From what to what? Why are you changing the subdomain? - there's no reason to .. all you need to do is use the ddns client to update the existing entry with the new ip address. Something's not adding up in your description.
Nick Howitt wrote:
You just have to remember that each version of PHP has its own php.ini file. I think you can also do the change in a .htaccess file, but I'd need to research it. If you can, that would work irrespective of the version of PHP.
I get where you're coming from, but htaccess doesn't appear to work - certainly not in a virtual website. I'd prefer to set it there as it's then only applicable to the individual website, but I really can't be ar*ed to try and figure out why it doesn't work! - changing it in php.ini works. - and there are other posts saying '.htaccess doesn't work' so I'm not prepared to waste any more time trying to figure out why; I've already wasted a couple of hours on it.
Tyler Randolph wrote:
Yes, it is using PHP 7.1 although doing that command that you mentioned still causes the message to show up: "The current PHP memory limit is below the recommended value of 512MB."
Is there another place that we would have to change memory_limit other than in /etc/php.ini?
That's all I changed (in the relevant /etc/opt/rh/php7.xx/php.ini)
FWIW, I'm running NC 20.0.5 installed from source. Moving to NC20, and subsequently a later version meant a couple of occ commands, but everything is working quite happily.
It's worth noting though, that NC 21 will need MariaDB version 10.2 or later - the default COS version is 5.5.65.
I had similar - but that improved by turning on the virtualization bits and adding a second processor. That makes it run faster (noticably).
The problem doesn't seem to be cpu (although I'll monitor that), it's definitely non-release of swap .. with a 2G swap partition, it starts minimal usage, then ramps up to over 1.5G used and not released. Odd. ClamAV shuts down, but the swap still isn't released.
Ok, I've a COS7 server that also acts as a VirtualBox host.
I've got two virtual machines running; a Windows 10 machine (no issues), and a COS7-based mail server.
Now, the Windows 10 VM doesn't appear to exhibit any problems, but the COS7 mail server seems to have a memory leak, that causes the swap partition (2G) to gradually fill until there's sod all free; the first symptom of which is that mail turns up 'unchecked' as ClamAV shuts down. Reboot the VM and everything works again.
I've got other linux vms that I run on my Windows 10 laptop (admittedly not continuously), that don't seem to exhibit the same problem. Anyone got any ideas?
I appreciate that it could be a virtualbox issue, but as the other windows VM doesn't have the problem, my initial thought is COS.
COS : 7.8.1 / Memory : 32G / Processor FX8350
VirtualBox : 6.1.14
COS : 7.8.1 / Memory : 2G
Yup. At the moment, I have to manually add it to the website site definition in flex-443 (I don't use the app; I get the original direct from NextCloud servers). NB: the redirect isn't necessary if ProxyPass is used as it already has the option to redirect http accesses to https.
I get that I could manually add a virtual host conf file (and create the various directories separately, but it would be much cleaner to leave the site definition/configuration at webconfig level and just have the option of enabling Transport Security at the same time - and as I say, if the period is definable, it makes the mechanism portable to any other site definition that may need it at a different rate.
Would it be possible to add Strict-Transport-Security to 443 definitions? (or at least the option to turn it on and define the period? It's noticeable that NextCloud installations (not sure about the app version) always complain that it's inactive whenever flex-443.conf is touched and it's a real bind having to keep adding it by hand!
Nick Howitt wrote:
Not so sure. I believe the app manipulates the firewall as necessary.
True, but it does also depend on where and how Port 80 is shut .. if it's just on the server, fine - but if it's closed (eg) on a seperate router, there's a problem .. and that's what keeps being forgotten - especially if the server is setup as standalone, with port forwarding carried out in a router!
There's a BIG difference between a COS system setup as a gateway and directly connected to the 'net, and a standalone with a seperate router and port forwarding! The question just never seems to get asked when someone complains that they can't get something to work from outside the local network.