[Terry A. Davis]

Man, I remember the last time I knew of an upgrade, it was something like a 32GB RAM server and that was considered blazing. That would've easily been over 10 years ago, crazy how time flies.

[Omniscient]

Did a fair amount of prep-work today to get this going.

Also creating scripts to make the php 7 -> php 8 upgrade easier.

If I'm lucky and things go well we can be on new server in a few weeks.

[Novo]

Is HF slow for you? When was the last time HF had a security breach? To be honest in a new setup it's just as much potential for less security as there is for more. All pages on HF should load very quickly. If not, LMK which ones don't. It is going to be improved faster hardware but when you're taking milliseconds in difference it's not noticeable.


RAM and CPU cycles. HF can run on a 250GB drive. Load on current server, as old as it is, less than 2 nearly all the time.

I also got an email saying provision will take 2 days. :(

We run software way heavier than any MyBB Forum on hardware from 2013, and its still going strong. Im not sure why the upgrade is needed.

If you think that your server needs high end servers, check out datapacket, maybe then you will be satisfied with the server.

[Omniscient]

We run software way heavier than any MyBB Forum on hardware from 2013, and its still going strong. Im not sure why the upgrade is needed.

If you think that your server needs high end servers, check out datapacket, maybe then you will be satisfied with the server.

We need the upgrade for moving forward with new and modern projects.

Example is if I want to add Voice Chat to Convo, I'll need to run new NPM libraries which aren't compatible with HF's old backend.

I also may want to do a github repo for HF where members can do pull requests. I have many members requesting this. They want to be able to build features but need to know HF's backend better.

[Novo]

We need the upgrade for moving forward with new and modern projects.

Example is if I want to add Voice Chat to Convo, I'll need to run new NPM libraries which aren't compatible with HF's old backend.

I also may want to do a github repo for HF where members can do pull requests. I have many members requesting this. They want to be able to build features but need to know HF's backend better.

Then you update the backend, not upgrade the entire server.

What are you even upgrading from and to?

[CL1B]

We need the upgrade for moving forward with new and modern projects.

Example is if I want to add Voice Chat to Convo, I'll need to run new NPM libraries which aren't compatible with HF's old backend.

I also may want to do a github repo for HF where members can do pull requests. I have many members requesting this. They want to be able to build features but need to know HF's backend better.

To be honest, Novo's perspective is many of our perspectives. There are many many many solutions that under HF's current load won't ever have a issue running it smoothly.

Maybe you could open up about how server infra looks in general so we as a community could help because quite frankly and for me personally i've always had the sense that you overcomplicate a lot of concepts. For example the current convo stack; you've mentioned it's a complex stack with Nodejs combined with php and a custom socket.io implementation and you were speaking of it as if it was really frustrating and complex for you but why not simplify it then?

A normal socket should be able to handle 1000 concurrent convo user at a time, chatting. On budget level hardware just fine. Really. When i say really I mean 1000%

There's also no reason to miss out on thousands of man hours of socket implementation especially for chat frameworks that you can find. The extreme one's can even handle more concurrent connections and messages p/s on even worse hardware!

And a heads up about voice chatting, not only does it require a whole different server that wouldn't work nice in a monolith server it's also barely doable if you do not want to get your study on about relaying and encoding/decoding WebRTC and encryption. Chances are high you will end up using a service like Cloudflare's TURN servers which is one of the most expensive services they have after the free tier, which if you know how it works; you will get there quick.

Genuinely I think these things should be reconsidered and maybe discussed broader, have you considered asking around on that forum you once showed to me? WebmasterTalk or Webtalk or something. I use reddit personally.

[Omniscient]

CL1B

Maybe you could open up about how server infra looks in general so we as a community could help because quite frankly and for me personally i've always had the sense that you overcomplicate a lot of concepts. For example the current convo stack; you've mentioned it's a complex stack with Nodejs combined with php and a custom socket.io implementation and you were speaking of it as if it was really frustrating and complex for you but why not simplify it then?

I didn't design the Convo feature, not at all. If I had, it would be completely different. I actually tend to simplify with possibly the exception of HF's security.

Going forward under a new server would allow me make some adjustments and changes to the structure but I'm still stuck in some ways.

MyBB is php, you can't get around that. I could build it all in JS but guess what, I'm not a JS developer, I don't even feel comfortable vibe coding the whole thing unless I understand the syntax well enough to guide AI properly on what I need. I also consider JS very limiting, insecure, and bloated. I would have to install a bunch of NPM's and dependencies. Very inefficient when PHP I don't ever really need libraries. Currently HF's PHP requires no dependencies like Composer. Everything is custom coded.

Convo is actually very feature rich for a custom chat solution for a single site. Instead members complain it doesn't have more features like Voice Chat.

It's not a custom socket.io solution, it's the old nodejs npm library. But it's a very old version (2.5) and current base is 4.8.1. Our NodeJS is v12 and new version is v24.19. We are about 6 years behind on version. That's signifcant with the past 2 years of vibe coding advances.

A normal socket should be able to handle 1000 concurrent convo user at a time, chatting. On budget level hardware just fine. Really. When i say really I mean 1000%

The resources aren't a problem. At least not until you start adding VC, then it's an issue. May even require a dedicated server.

And a heads up about voice chatting, not only does it require a whole different server that wouldn't work nice in a monolith server it's also barely doable if you do not want to get your study on about relaying and encoding/decoding WebRTC and encryption.

I've already done the research. Even created a PoC a while ago.

https://hackforums.net/webrtc/test.html

But to really move forward, we must be on a new server. Vibe coding AI cannot deal with the outdated server well.

Genuinely I think these things should be reconsidered and maybe discussed broader

I like discussing things with the community but overall, it's not required. I am the one who does all the work.

[Dotexe]

To be honest I support Omni's move, we should upgrade in order to receive new stuff, really depends on the feature itself, most of them can be coded in older PHP version but that'd require more and more code and at some point if Omni decides to vibe code it to faster the process, that will only lead to problems in the future, especially from security posture. There's plenty of people here who can test the security and report something to Omni, of course all of us want a safer place for everyone and Omni himself can't do everything alone.

[CL1B]

CL1B


I didn't design the Convo feature, not at all. If I had, it would be completely different. I actually tend to simplify with possibly the exception of HF's security.

Going forward under a new server would allow me make some adjustments and changes to the structure but I'm still stuck in some ways.

MyBB is php, you can't get around that. I could build it all in JS but guess what, I'm not a JS developer, I don't even feel comfortable vibe coding the whole thing unless I understand the syntax well enough to guide AI properly on what I need. I also consider JS very limiting, insecure, and bloated. I would have to install a bunch of NPM's and dependencies. Very inefficient when PHP I don't ever really need libraries. Currently HF's PHP requires no dependencies like Composer. Everything is custom coded.

Convo is actually very feature rich for a custom chat solution for a single site. Instead members complain it doesn't have more features like Voice Chat.

It's not a custom socket.io solution, it's the old nodejs npm library. But it's a very old version (2.5) and current base is 4.8.1. Our NodeJS is v12 and new version is v24.19. We are about 6 years behind on version. That's signifcant with the past 2 years of vibe coding advances.


The resources aren't a problem. At least not until you start adding VC, then it's an issue. May even require a dedicated server.


I've already done the research. Even created a PoC a while ago.

https://hackforums.net/webrtc/test.html

But to really move forward, we must be on a new server. Vibe coding AI cannot deal with the outdated server well.


I like discussing things with the community but overall, it's not required. I am the one who does all the work.

Yeah, well.. All we can do is wish you the best of luck with it. BTW, I wouldn't use any productional servers for vibe coding ever. Instead I'd recommend you get that dev branch running locally or on a dev server. You don't need GH for it. Also as said in my previous post please be aware of the costs related to voice chatting.. say the current convo community would've used VC (only the active ones) you are almost certainly looking at hundreds of dollars for that small stack a month; with no monthly monetization stream this will dry you up quicker than anything else.

Iirc this is the reason why most discord alternatives even open source ones fail and why discord needed such a large angel investment to begin with.

[Omniscient]

Yeah, well.. All we can do is wish you the best of luck with it. BTW, I wouldn't use any productional servers for vibe coding ever. Instead I'd recommend you get that dev branch running locally or on a dev server. You don't need GH for it. Also as said in my previous post please be aware of the costs related to voice chatting.. say the current convo community would've used VC (only the active ones) you are almost certainly looking at hundreds of dollars for that small stack a month; with no monthly monetization stream this will dry you up quicker than anything else.

Iirc this is the reason why most discord alternatives even open source ones fail and why discord needed such a large angel investment to begin with.

I am always very cost-minded. You don't have to worry about that.

And I do a ton of stuff on production server but I feel you about vibe coding on it. I have a plan. Don't worry about it.

[Dystortion]

Iirc this is the reason why most discord alternatives even open source ones fail and why discord needed such a large angel investment to begin with.

I see what you're saying here but it's actually quite inexpensive if you build it right. pricing you're looking at CF - $0.05/GB egress, with a 1,000 GB free tier. Upload traffic into the SFU is not charged, and TURN traffic used alongside the SFU isn't double-charged

Let's run with Opus as the budget options roughly 16–20 kbps for wideband and 28–40 kbps for full-band.

32 kbps Opus + 25% allowance for transport overhead (approx)= 40 kbps.

Roughly 18 MB per hour per forwarded speaker/listener stream. Now assume an average of 1.5 people actually speaking simultaneously, because humans generally haven't yet evolved the ability to have 37 comprehensible conversations at once.

People sitting in voice Approx egress/hour 24/7 equivalent/month
10 (approx)=10.24 GB (approx)=175 GB
50 (approx)=11.32 GB (approx)=1953 GB
100 (approx)=12.67 GB (approx)=11.92 TB

Theoretically a 50-person voice room could run continuously all month and still land around the free-tier boundary. 100 people permanently connected 24/7 comes to about 1.9 TB under this rough model. First 1TB free. Remaining is like $46/month.

[Omniscient]

HF New Server Setup Progress

I've started building out a new dedicated server that will eventually become the new home for Hack Forums.

This is being done as a fresh build rather than trying to upgrade the current production server in place. The goal is to get everything installed, hardened, configured, migrated, and extensively tested before any production DNS changes are made.

Hardware

The new server is significantly more capable than the existing environment:

• AMD EPYC server CPU
  
• 256GB ECC RAM
  
• Enterprise SSD storage in hardware RAID-1
  
• 1Gbps dedicated uplink
  
• Large monthly bandwidth allocation
  
• Datacenter-level DDoS protection
  
• Out-of-band remote management
  

Operating System / Control Panel

• Fresh Debian installation
  
• Nginx frontend with Apache backend
  
• PHP-FPM
  
• MariaDB
  
• Fail2ban
  

Initial Server Hardening

So far I've:

• Changed and hardened SSH configuration
  
• Configured SSH key authentication
  
• Set up firewall rules
  
• Added administrative IP whitelisting
  
• Disabled unnecessary public services
  
• Disabled public ICMP echo responses
  
• Configured Fail2ban
  
• Restricted various administrative services
  
• Configured a dedicated server hostname
  

I'm intentionally not posting IP addresses, ports, whitelist information, or the detailed firewall layout.

PHP

The server is currently running PHP 8.2 through PHP-FPM.

I've verified the PHP environment contains everything HF should need, including:

• mysqli / mysqlnd
  
• mbstring
  
• curl
  
• GD
  
• ImageMagick
  
• intl
  
• OpenSSL
  
• sodium
  
• XML / DOM / SimpleXML
  
• ZIP
  
• SOAP
  
• sockets
  
• APCu
  
• Memcached
  
• OPcache
  

GD was also checked and has JPEG, PNG, GIF, WebP, AVIF, FreeType and other modern image format support.

The PHP-FPM configuration already contained a fairly strong disabled-functions policy. Dangerous web-facing process execution functions such as shell execution, exec, proc_open, popen and PCNTL process controls are disabled.

MariaDB

MariaDB is installed.

I've created a custom database configuration specifically for HF rather than using the generic hosting defaults.

Some of the initial tuning includes:

• 96GB InnoDB buffer pool
  
• Larger InnoDB redo/log buffers
  
• Increased connection capacity
  
• Increased table caches
  
• Increased thread cache
  
• Larger temporary table limits
  
• SSD-oriented InnoDB I/O settings
  
• Query cache disabled
  
• DNS resolution disabled for DB connections
  
• Slow query logging enabled
  
• Increased file descriptor limits
  

The database configuration is only a baseline. Once a realistic copy of the HF database is imported and we're generating actual traffic, I'll tune it further based on real performance statistics instead of guessing.

SSL / Testing

Since production DNS still points at the current HF server, I've configured the new environment so I can test domains locally without changing public DNS.

Temporary SSL certificates are installed for development/testing, so I can test the site using its real hostname while everyone else continues using the existing production server.

Current Status

At this point the underlying operating system, web stack, PHP environment, database server, SSL, firewall and basic server security are in place.

The next major stage is bringing over a copy of the HF database and codebase and beginning actual compatibility and performance testing.

A major part of this migration will also be moving the old PHP 7-era HF codebase forward to PHP 8.2 compatibility. That will be done entirely on the new/test environment before anything reaches production.

There is still a lot of work left, but the base server is now coming together.

[Terry A. Davis]

Would you consider opening a "Demo" version of HF for users to trial on an invite basis? Even just to view not to edit, just to see if we can spot any bugs or leaks in it before you fully go live?

[Omniscient]

Would you consider opening a "Demo" version of HF for users to trial on an invite basis? Even just to view not to edit, just to see if we can spot any bugs or leaks in it before you fully go live?

No, because of how I'm going to dev this. For me to do the work I have to custom DNS entry for the domain. Dev site has to be under hackforums.net because of all the hard-coded features using the direct domain. SSL doesn't even work and I have to allow the insecure connection. Some browsers won't even let you connect to http anymore.

When I officially change HF's DNS to the new server, everyone can start testing what's broken then. I'm sure things will be broken too. It's not optimal but it's more efficient.

Today I have at least two things to complete for the new server (Tunnel and Websocket proxy). Should not take me too long and then I believe I'll be ready for site move (db and files). DB installation is going to take hours alone.

Now that summer is nearing an end, I have the time to focus on finishing this once and for all.

[Transporter007]

HF New Server Setup Progress

I've started building out a new dedicated server that will eventually become the new home for Hack Forums.

This is being done as a fresh build rather than trying to upgrade the current production server in place. The goal is to get everything installed, hardened, configured, migrated, and extensively tested before any production DNS changes are made.

Hardware

The new server is significantly more capable than the existing environment:

• AMD EPYC server CPU
  
• 256GB ECC RAM
  
• Enterprise SSD storage in hardware RAID-1
  
• 1Gbps dedicated uplink
  
• Large monthly bandwidth allocation
  
• Datacenter-level DDoS protection
  
• Out-of-band remote management
  

Operating System / Control Panel

• Fresh Debian installation
  
• Nginx frontend with Apache backend
  
• PHP-FPM
  
• MariaDB
  
• Fail2ban
  

Initial Server Hardening

So far I've:

• Changed and hardened SSH configuration
  
• Configured SSH key authentication
  
• Set up firewall rules
  
• Added administrative IP whitelisting
  
• Disabled unnecessary public services
  
• Disabled public ICMP echo responses
  
• Configured Fail2ban
  
• Restricted various administrative services
  
• Configured a dedicated server hostname
  

I'm intentionally not posting IP addresses, ports, whitelist information, or the detailed firewall layout.

PHP

The server is currently running PHP 8.2 through PHP-FPM.

I've verified the PHP environment contains everything HF should need, including:

• mysqli / mysqlnd
  
• mbstring
  
• curl
  
• GD
  
• ImageMagick
  
• intl
  
• OpenSSL
  
• sodium
  
• XML / DOM / SimpleXML
  
• ZIP
  
• SOAP
  
• sockets
  
• APCu
  
• Memcached
  
• OPcache
  

GD was also checked and has JPEG, PNG, GIF, WebP, AVIF, FreeType and other modern image format support.

The PHP-FPM configuration already contained a fairly strong disabled-functions policy. Dangerous web-facing process execution functions such as shell execution, exec, proc_open, popen and PCNTL process controls are disabled.

MariaDB

MariaDB is installed.

I've created a custom database configuration specifically for HF rather than using the generic hosting defaults.

Some of the initial tuning includes:

• 96GB InnoDB buffer pool
  
• Larger InnoDB redo/log buffers
  
• Increased connection capacity
  
• Increased table caches
  
• Increased thread cache
  
• Larger temporary table limits
  
• SSD-oriented InnoDB I/O settings
  
• Query cache disabled
  
• DNS resolution disabled for DB connections
  
• Slow query logging enabled
  
• Increased file descriptor limits
  

The database configuration is only a baseline. Once a realistic copy of the HF database is imported and we're generating actual traffic, I'll tune it further based on real performance statistics instead of guessing.

SSL / Testing

Since production DNS still points at the current HF server, I've configured the new environment so I can test domains locally without changing public DNS.

Temporary SSL certificates are installed for development/testing, so I can test the site using its real hostname while everyone else continues using the existing production server.

Current Status

At this point the underlying operating system, web stack, PHP environment, database server, SSL, firewall and basic server security are in place.

The next major stage is bringing over a copy of the HF database and codebase and beginning actual compatibility and performance testing.

A major part of this migration will also be moving the old PHP 7-era HF codebase forward to PHP 8.2 compatibility. That will be done entirely on the new/test environment before anything reaches production.

There is still a lot of work left, but the base server is now coming together.

Whats the cost of it per month ?