29-09-2024, 08:41 PM
TLDR: Server go burrrr, updates & upgrades make life better. Backups ftw.
As of a way of an update to this. Although i doubt anyone cares.
We are now running on a proxmox cluster (instead of vMware.. ahem broadcom), we shall just causally ignore the various server OS version upgrades, mostly because they are (as they should be) transparent to the consumer which requires said functionality and security. More importantly, something we have had for a while now, a dedicated backup server. All working hunky dory.
However, my amazing idea of putting hardware raid in each of my servers and then running proxmox was short sighted, because promox runs ZFS which is a superior SOFTWARE raid system, rather than the hardware one i "upgraded to". Short story, they dont play nice.
The problem was on the first upgrade of the new host OS that, as i mentioned, required CPU upgrades on. (luckily secondary server was still acting as main host. Because best practice is always upgrade your standby node first, glad i did!!) died because of general corruption. I have this server in my office right now, i took the opportunity to upgrade the CPU's and remove the hardware raid for the main storage area (still raid1 for main OS).
I've yet to get enough time to put it back in the rack, but when i do it will be re-added to the failover cluster and become the main node, whilst i make "adjustments" to the current primary node (which is actually the secondary/failover node) Which, i've not updated because i fear for the same fate as my current secondary node, which is currently primary. I'm going to give up with the node specification type as im sure its confusing and boring, but please do have faith. For infrastructure reasons they are both the same model and spec servers, which means they are both running hardware raid, therefore, i dare not upgrade it. Because i peformed the same actions to both servers. Given that one of them binned it on the first upgrade, untill i can understand and fix the first node (or server) then i shouldnt perform any further upgrades.
Current plan is to failover to this one in my office (becoming primary), then rebuild 2nd host (current primary host) to be the same as new primary. Ensure backups keep running, maybe install and setup CEPH (automatic replication of data without a 3rd server acting as storage host) and see if i can actually get automatic failover working so i can ignore this server for the rest of my life haha
Live long and 306oc. No, that wasnt it. Living for the 306!
Sorry its taken me nearly a year to actually address this, i never actually noticed this comment till i read back on this thread. We can totally do this, admin approval on signups etc, that's a valid option. The challange is, to avoid spam and troll signups, it requires either a human or a technology solution (some of which are already implemented, but agreed, poorly maintained). Therefore it requires human intervention or an automation solution based on modern AI. Although AI has got a lot better over recent years, i've have not yet learned or seen anything that makes a current LLM/AI which is appropriate for open source forums, like this one. The point is, having an admin approval step destroys user experiance (in my opinion) and if it requires manual approval, it causes delay and requires activity from the people that can approve. Its a careful balance of allowing people into the trusted community vs stopping bots/trolls etc
You have to remember though, on the balance of engagement over ease of use. Its a signifcantly bigger topic than it was when this forum started. There is a lot to break down in that statement, but from a visibility perspective it has to make sense to continue. Volume if users doesnt mean quality or profit. Its especially important in the environement we operate, the costs only scale with more users (not that i care and the way i've structured the forum, they actually dont, most use cases do have cost scaling though).
Given how stale the forum is, and likely will continue to be, unless we could be more encompassing of other market drivers. E.G the "only peugeot forum". Its just not going to happen, there are better, more organised clubs than us that focus on the peugeot experiance and if thats what you want, please go join their forum. There is no competitioncompetmpehere, it should be about supporting each other.
My plan, for a long time was to incorporate Single Sign On. Eg, people could use their facebook/google/apple account etc to login to the forum so it was easy.
Unfortunality at the time the technology to do this was immature and i didnt (still dont really) posess the skills to incroprate SSO into a forum that doesnt natively support multiple signup options, of the SSO/SAML nature.
I know some of the paid forums have this functionality, but having a forum i have to pay a license/subscription fee for (whilst, it might improve user experience) wasnt the goal of 306oc. And by this i mean, it was to give the users the best experiance at the lowest cost possible. Rather than just throwing a bunch of money at fully devleoped forum software. Our goal was to have an amazing community, share knoweldge and our passion for 306's in a way that people could enjoy and support, without having to open their wallet. Knoweldge should be free, a platform shouldnt be make money from other peopels skills.
Therefore anything that has mandatory future cost that means you have to think strategically and look at the ROI of the platform you are running isnt viable. We made the choice as an admin team to ensure we kept operating costs as sensible as possible to guarantee longevity.
Our goal was to be able to provide free and open access to information that could be gathered and we had to ensure our running costs where as sensible as possible, meaning that this effort of people posting using information could be free and accessable.
The forum software (to my knoweldge) still doesnt support SSO natively. So i'd have to get somebody with the programming and API skills to do it. Its just not worth the risk to do this, security wise anyway.
They've been promising HTML5/SSO Ready/OMG SHINY version 2.0 of MYBB for what feels like 10 years and its still not here. At the point the software supports it without my effort i'm all on board. However that itself brings different security/spam problems and if i refer back to the goal of the forum, i dont want it full of bollocks if i can avoid it.
As of a way of an update to this. Although i doubt anyone cares.
We are now running on a proxmox cluster (instead of vMware.. ahem broadcom), we shall just causally ignore the various server OS version upgrades, mostly because they are (as they should be) transparent to the consumer which requires said functionality and security. More importantly, something we have had for a while now, a dedicated backup server. All working hunky dory.
However, my amazing idea of putting hardware raid in each of my servers and then running proxmox was short sighted, because promox runs ZFS which is a superior SOFTWARE raid system, rather than the hardware one i "upgraded to". Short story, they dont play nice.
The problem was on the first upgrade of the new host OS that, as i mentioned, required CPU upgrades on. (luckily secondary server was still acting as main host. Because best practice is always upgrade your standby node first, glad i did!!) died because of general corruption. I have this server in my office right now, i took the opportunity to upgrade the CPU's and remove the hardware raid for the main storage area (still raid1 for main OS).
I've yet to get enough time to put it back in the rack, but when i do it will be re-added to the failover cluster and become the main node, whilst i make "adjustments" to the current primary node (which is actually the secondary/failover node) Which, i've not updated because i fear for the same fate as my current secondary node, which is currently primary. I'm going to give up with the node specification type as im sure its confusing and boring, but please do have faith. For infrastructure reasons they are both the same model and spec servers, which means they are both running hardware raid, therefore, i dare not upgrade it. Because i peformed the same actions to both servers. Given that one of them binned it on the first upgrade, untill i can understand and fix the first node (or server) then i shouldnt perform any further upgrades.
Current plan is to failover to this one in my office (becoming primary), then rebuild 2nd host (current primary host) to be the same as new primary. Ensure backups keep running, maybe install and setup CEPH (automatic replication of data without a 3rd server acting as storage host) and see if i can actually get automatic failover working so i can ignore this server for the rest of my life haha
Live long and 306oc. No, that wasnt it. Living for the 306!
(02-12-2023, 01:27 PM)Mighty306 Wrote: I've been wondering about a way to allow new members again. Set up an email address that people have to contact & then use admin priviledges to set up an account on their behalf?
Sorry its taken me nearly a year to actually address this, i never actually noticed this comment till i read back on this thread. We can totally do this, admin approval on signups etc, that's a valid option. The challange is, to avoid spam and troll signups, it requires either a human or a technology solution (some of which are already implemented, but agreed, poorly maintained). Therefore it requires human intervention or an automation solution based on modern AI. Although AI has got a lot better over recent years, i've have not yet learned or seen anything that makes a current LLM/AI which is appropriate for open source forums, like this one. The point is, having an admin approval step destroys user experiance (in my opinion) and if it requires manual approval, it causes delay and requires activity from the people that can approve. Its a careful balance of allowing people into the trusted community vs stopping bots/trolls etc
You have to remember though, on the balance of engagement over ease of use. Its a signifcantly bigger topic than it was when this forum started. There is a lot to break down in that statement, but from a visibility perspective it has to make sense to continue. Volume if users doesnt mean quality or profit. Its especially important in the environement we operate, the costs only scale with more users (not that i care and the way i've structured the forum, they actually dont, most use cases do have cost scaling though).
Given how stale the forum is, and likely will continue to be, unless we could be more encompassing of other market drivers. E.G the "only peugeot forum". Its just not going to happen, there are better, more organised clubs than us that focus on the peugeot experiance and if thats what you want, please go join their forum. There is no competitioncompetmpehere, it should be about supporting each other.
My plan, for a long time was to incorporate Single Sign On. Eg, people could use their facebook/google/apple account etc to login to the forum so it was easy.
Unfortunality at the time the technology to do this was immature and i didnt (still dont really) posess the skills to incroprate SSO into a forum that doesnt natively support multiple signup options, of the SSO/SAML nature.
I know some of the paid forums have this functionality, but having a forum i have to pay a license/subscription fee for (whilst, it might improve user experience) wasnt the goal of 306oc. And by this i mean, it was to give the users the best experiance at the lowest cost possible. Rather than just throwing a bunch of money at fully devleoped forum software. Our goal was to have an amazing community, share knoweldge and our passion for 306's in a way that people could enjoy and support, without having to open their wallet. Knoweldge should be free, a platform shouldnt be make money from other peopels skills.
Therefore anything that has mandatory future cost that means you have to think strategically and look at the ROI of the platform you are running isnt viable. We made the choice as an admin team to ensure we kept operating costs as sensible as possible to guarantee longevity.
Our goal was to be able to provide free and open access to information that could be gathered and we had to ensure our running costs where as sensible as possible, meaning that this effort of people posting using information could be free and accessable.
The forum software (to my knoweldge) still doesnt support SSO natively. So i'd have to get somebody with the programming and API skills to do it. Its just not worth the risk to do this, security wise anyway.
They've been promising HTML5/SSO Ready/OMG SHINY version 2.0 of MYBB for what feels like 10 years and its still not here. At the point the software supports it without my effort i'm all on board. However that itself brings different security/spam problems and if i refer back to the goal of the forum, i dont want it full of bollocks if i can avoid it.
Team - Inappropriate Footwear. Member 1 of 3.
1. Fooby's Car
2. Fooby's Hire Car(s)
3. Fooby's Company Pool Car
Team Self Improvement
1. Learn to Car
1. Fooby's Car
2. Fooby's Hire Car(s)
3. Fooby's Company Pool Car
Team Self Improvement
1. Learn to Car