SlickStack on the 512MB RAM cloud server?

pateldev

From their homepage it recommended 1GB RAM or more for Ubuntu/Mysql stability. On the Discord they said it was never tested using lesser RAM memory, but now includes 2GB swapfile since 2021 that will possibly help improve stability on the low memory VPS servers.

did anyone try SS on the 512MB server, and your recommend to improve stability? Cuz if I can launch 100 websites on low memory it will be like $2.50 only each which is amazing (not for Woocommerce)

kinda surprised nobody is talking about WordPress stack scripts here I see only one 2019 topic

talk.lowendspirit.com/discussion/336/anyone-tested-easyengine

Mr_Tom

I don't see why it wouldn't work with some swap space.

If the $2.50 is per month, then that's still $25/month for the 100 servers. You'd get a better spec machine for less than that which would run all 100 sites in one server.

chocolateshirt

Well, you could use Webinoly on 512 MB Box, it work great and it support multiple domain/site.

pateldev

@Mr_Tom said:
I don't see why it wouldn't work with some swap space.

If the $2.50 is per month, then that's still $25/month for the 100 servers. You'd get a better spec machine for less than that which would run all 100 sites in one server.

I think you mean $250/month but I understand your meaning. I don't need much storage for each WP site, is it a bad idea to make the swapfile really big in SlickStack like 10GB or something big like that? or will make problems

@chocolateshirt said:
Well, you could use Webinoly on 512 MB Box, it work great and it support multiple domain/site.

thanks I read about Webinoly also seems cool but mostly I always putting my client on a single VPS for better SEO results and if any problem only one customer is angry and not 100 customers phoning me lol

or maybe it can be like "Webinoly Japan VPS" and "Webinoly Germany VPS" or something can help divide it a little

the problem is sometimes my customer needs to upgrade the server during traffic growth (or disk space) so easier to keep some of "VIP" customers on the single VPS if upgrade will be required later

chocolateshirt

@pateldev said:

@chocolateshirt said:
Well, you could use Webinoly on 512 MB Box, it work great and it support multiple domain/site.

thanks I read about Webinoly also seems cool but mostly I always putting my client on a single VPS for better SEO results and if any problem only one customer is angry and not 100 customers phoning me lol

Do you mean one web site per IPv4 address?

Mr_Tom

@pateldev said: I think you mean $250/month but I understand your meaning. I don't need much storage for each WP site, is it a bad idea to make the swapfile really big in Ubuntu like 10GB or something big like that? or will make problems

Yeah, apologies I didn't maths very well there lol.

On a virtual server you'll likely hit IO issues if everything is running in swap. While drives are getting faster (NVMe etc) it's still slower than using the RAM itself.

If the sites aren't woocommerce or very complicated then you should be able to run all 100 sites off a 2gb/4gb RAM VM - if they're not optimised enough to run like that then they'll not run well on their own 512mb server - but even if that is the case you'll still be under budget for something that should run them all.

ehab

@Ympker and @vyas are the wordpress and alike kings .. @vyas is actually a beast on this topic so they should say something.

pateldev

@chocolateshirt said:

@pateldev said:

@chocolateshirt said:
Well, you could use Webinoly on 512 MB Box, it work great and it support multiple domain/site.

thanks I read about Webinoly also seems cool but mostly I always putting my client on a single VPS for better SEO results and if any problem only one customer is angry and not 100 customers phoning me lol

Do you mean one web site per IPv4 address?

Yes but seems dedicated IP address is not so important these days, however local speed is helping a lot. Like for a local business in ex. Dubai it helps a lot to host in the single VPS in Dubai and not sharing the server in USA (in my experience) and for SlickStack it will also include HSTS and security headers which seems to be secondary ranking factors in Google now

@Mr_Tom said:

@pateldev said: I think you mean $250/month but I understand your meaning. I don't need much storage for each WP site, is it a bad idea to make the swapfile really big in Ubuntu like 10GB or something big like that? or will make problems

Yeah, apologies I didn't maths very well there lol.

On a virtual server you'll likely hit IO issues if everything is running in swap. While drives are getting faster (NVMe etc) it's still slower than using the RAM itself.

If the sites aren't woocommerce or very complicated then you should be able to run all 100 sites off a 2gb/4gb RAM VM - if they're not optimised enough to run like that then they'll not run well on their own 512mb server - but even if that is the case you'll still be under budget for something that should run them all.

hmm interesting this is what I want to learn... cuz I noticed in the 1GB memory server I check free -m and seems the OS is using 2GB swapfile ALWAYS even if there is available RAM memory and I don't know why, must reboot sometimes and the usage of swapfile will decrease after reboot, but maybe that is no problem really just allow Linux to decide?

seems Mysql is the main problem for relying on swapfile because Cloudflare/CDN will be loading the static files anyway?

Mr_Tom

@pateldev said: hmm interesting this is what I want to learn... cuz I noticed in the 1GB memory server I check free -m and seems the OS is using 2GB swapfile ALWAYS even if there is available RAM memory and I don't know why

Sounds like something isn't configured correctly - what's the server doing?

If it's idle then it should barely be using the RAM, let alone swap.

# free -h
               total        used        free      shared  buff/cache   available
Mem:           976Mi        70Mi       251Mi       0.0Ki       654Mi       756Mi
Swap:          1.0Gi       2.0Mi       1.0Gi

If it's an active server with active site, how many requests and what type of site? Things such as PHP/MySQL/CMS version I find can all make small differences to usage - I moved a Magento site from version 2.1 to 2.4 recently and ram usage dropped about 5gb.

pateldev

@Mr_Tom said:

@pateldev said: hmm interesting this is what I want to learn... cuz I noticed in the 1GB memory server I check free -m and seems the OS is using 2GB swapfile ALWAYS even if there is available RAM memory and I don't know why

Sounds like something isn't configured correctly - what's the server doing?

If it's idle then it should barely be using the RAM, let alone swap.

# free -h
total used free shared buff/cache available
Mem: 976Mi 70Mi 251Mi 0.0Ki 654Mi 756Mi
Swap: 1.0Gi 2.0Mi 1.0Gi

If it's an active server with active site, how many requests and what type of site? Things such as PHP/MySQL/CMS version I find can all make small differences to usage - I moved a Magento site from version 2.1 to 2.4 recently and ram usage dropped about 5gb.

yes sorry my observation is from the active website, I believed swap is used after RAM is 100% full but sometimes it seems like swapfile is being used but RAM is only 75% full or something but maybe just Linux algorithm

is swapfile on modern SSD drives pretty fast? KVM and SSD hard drive can help 512MB RAM server scaling better? if the site is a tiny bit slower for complex Mysql queries (only) maybe it's okay for me

seems nobody is using the cheap VPS for WordPress (SEO sites) here or you guys mostly using static HTML sites?

vyas

@chocolateshirt said:
Well, you could use Webinoly on 512 MB Box, it work great and it support multiple domain/site.

Webinoly + Debian + upto 4 WP sites runs absolutely fine on 256 MB Ram

@pateldev said:
seems nobody is using the cheap VPS for WordPress (SEO sites) here or you guys mostly using static HTML sites?

I have a producton site on 512 MB LittleVZ. Depends on how much time/ effort one aspires to spend on optimizations

Mr_Tom

Wordpress on a small server is fine just make sure its optimised well and itll be quick.

I’d use a LEMP stack and adjust everything to suit. If theyre mainly content pages you can make wordpress fast even without caching plugins.

flips

@pateldev said: kinda surprised nobody is talking about WordPress stack scripts here I see only one 2019 topic
https://talk.lowendspirit.com/discussion/336/anyone-tested-easyengine

Ah, my old thread ... I believe the topic has been discussed more, maybe in the Wordpress thread or in an older version of the Pit ... or maybe it was on HostedTalk ...

vyas

@Mr_Tom said:
Wordpress on a small server is fine just make sure its optimised well and itll be quick.

I’d use a LEMP stack and adjust everything to suit. If theyre mainly content pages you can make wordpress fast even without caching plugins.

Very succintly put.
@yoursunny may make arguments in favour of caddy + wp, with higher ram redis might also do the trick

johnk

hmm interesting this is what I want to learn... cuz I noticed in the 1GB memory server I check free -m and seems the OS is using 2GB swapfile ALWAYS even if there is available RAM memory and I don't know why, must reboot sometimes and the usage of swapfile will decrease after reboot, but maybe that is no problem really just allow Linux to decide?

vm.swapiness is what you want to take a look at. Swap, contrary to public belief, isn't just used for when RAM runs out. There are a variety of things that determine memory pressure, and things like page-size play a role too.

Yes but seems dedicated IP address is not so important these days, however local speed is helping a lot. Like for a local business in ex. Dubai it helps a lot to host in the single VPS in Dubai and not sharing the server in USA (in my experience) and for SlickStack it will also include HSTS and security headers which seems to be secondary ranking factors in Google now

A dedicated IP is irrelevant, makes zero difference. Yes, local performance matters, but I'd consider just 3-4 VPSes dispersed locally. HSTS/CSP are not Google ranking factors.

I think you mean $250/month but I understand your meaning. I don't need much storage for each WP site, is it a bad idea to make the swapfile really big in SlickStack like 10GB or something big like that? or will make problems

It's important to understand where memory is used in a LEMP/LAMP/whatever stack. PHP uses memory for caching - opcache, realpath cache, etc. MySQL buffers/tmp tables are in memory. NGINX/Apache may cache file descriptors in memory. If you still having enough, the kernel will cache files in memory. All these are because these are latency-sensitive operations. Ie, you're talking 0.5ms to 1 ms via disk, vs many magnitudes lower in memory. Obviously, you can configure LEMP to do everything in disk. It'll arguably be faster than having to write everything to a swap file first. But ultimately, swap is not a memory replacement, and depending on how your stack is configure, you will see minimally improved, if not worse, performance by allocating everything to swap.

FWIW, I really would not run WordPress on anything <1 GB. At 256/512, you really don't have ample for proper caching, and you leave practically no overhead for run away processes, or traffic surges.

pateldev

Yes I know about LEMP stack and FastCGI cache (Nginx) that is what SlickStack used already.

"just make sure your settings are good" yes that is the reason for my topic cuz I want to know if anyone tried to use SlickStack on 512MB or 256MB RAM yet? or if default SS settings can be adjusted better in the ss-config file for this purpose

@vyas said:

@chocolateshirt said:
Well, you could use Webinoly on 512 MB Box, it work great and it support multiple domain/site.

Webinoly + Debian + upto 4 WP sites runs absolutely fine on 256 MB Ram

Interesting maybe Webinoly used less RAM memory by default but truly not sure, cuz I never compare with SlickStack

@johnk said:

hmm interesting this is what I want to learn... cuz I noticed in the 1GB memory server I check free -m and seems the OS is using 2GB swapfile ALWAYS even if there is available RAM memory and I don't know why, must reboot sometimes and the usage of swapfile will decrease after reboot, but maybe that is no problem really just allow Linux to decide?

vm.swapiness is what you want to take a look at. Swap, contrary to public belief, isn't just used for when RAM runs out. There are a variety of things that determine memory pressure, and things like page-size play a role too.

ahh.... great info... but seems not available in the ss-config options
https://github.com/littlebizzy/slickstack/blob/master/bash/ss-config-sample.txt

Yes but seems dedicated IP address is not so important these days, however local speed is helping a lot. Like for a local business in ex. Dubai it helps a lot to host in the single VPS in Dubai and not sharing the server in USA (in my experience) and for SlickStack it will also include HSTS and security headers which seems to be secondary ranking factors in Google now

A dedicated IP is irrelevant, makes zero difference. Yes, local performance matters, but I'd consider just 3-4 VPSes dispersed locally. HSTS/CSP are not Google ranking factors.

not sure, well I outrank many SERPs using these and seems not much else different from competitor sites so maybe not the truly reason but I will not be surprised if HSTS is secondary factor since SSL is a factor already

I think you mean $250/month but I understand your meaning. I don't need much storage for each WP site, is it a bad idea to make the swapfile really big in SlickStack like 10GB or something big like that? or will make problems

It's important to understand where memory is used in a LEMP/LAMP/whatever stack. PHP uses memory for caching - opcache, realpath cache, etc. MySQL buffers/tmp tables are in memory. NGINX/Apache may cache file descriptors in memory. If you still having enough, the kernel will cache files in memory. All these are because these are latency-sensitive operations. Ie, you're talking 0.5ms to 1 ms via disk, vs many magnitudes lower in memory. Obviously, you can configure LEMP to do everything in disk. It'll arguably be faster than having to write everything to a swap file first. But ultimately, swap is not a memory replacement, and depending on how your stack is configure, you will see minimally improved, if not worse, performance by allocating everything to swap.

FWIW, I really would not run WordPress on anything <1 GB. At 256/512, you really don't have ample for proper caching, and you leave practically no overhead for run away processes, or traffic surges.

REALLY good info..... THANKS..... anyway I never tried it before but mostly want to know if possible

yoursunny

@vyas said:
@yoursunny may make arguments in favour of caddy + wp, with higher ram redis might also do the trick

@yoursunny will #DeleteWordPress and use a static site generator.

johnk

@johnk said:

hmm interesting this is what I want to learn... cuz I noticed in the 1GB memory server I check free -m and seems the OS is using 2GB swapfile ALWAYS even if there is available RAM memory and I don't know why, must reboot sometimes and the usage of swapfile will decrease after reboot, but maybe that is no problem really just allow Linux to decide?

vm.swapiness is what you want to take a look at. Swap, contrary to public belief, isn't just used for when RAM runs out. There are a variety of things that determine memory pressure, and things like page-size play a role too.

ahh.... great info... but seems not available in the ss-config options
https://github.com/littlebizzy/slickstack/blob/master/bash/ss-config-sample.txt

It's likely not a slickstack config option. It's a kernel parameter you can tune via editing /etc/sysctl.conf / /etc/sysctl.d/.conf / sysctl -w "vm.swapiness=0-100"

Yes but seems dedicated IP address is not so important these days, however local speed is helping a lot. Like for a local business in ex. Dubai it helps a lot to host in the single VPS in Dubai and not sharing the server in USA (in my experience) and for SlickStack it will also include HSTS and security headers which seems to be secondary ranking factors in Google now

A dedicated IP is irrelevant, makes zero difference. Yes, local performance matters, but I'd consider just 3-4 VPSes dispersed locally. HSTS/CSP are not Google ranking factors.

not sure, well I outrank many SERPs using these and seems not much else different from competitor sites so maybe not the truly reason but I will not be surprised if HSTS is secondary factor since SSL is a factor already

>

Correlation != causation, especially with something as complex as Google. SSL is a factor, because it makes a very big difference in security, and in this day and age, there's no reason not to use SSL. The benefits of HSTS are nowhere near those of HSTS, and make little difference on the quality of websites. Remember, the purpose of Google is to provide the user with useful and relevant information.

Signals like CWV/SSL are included, because they contribute to user experience. HSTS does not, and therefore Google has no motivation to introduce arbitrary data that could potentially dilute their results.

havoc

There is also a real question as to whether it is worth the hassle in terms of value of your time. Depending on country even an hour of additional tune at minimum wage could probably cover a more expensive vps