Hosting Issues

OYP was sporadically down for short amounts of time, and recently it was down for about two days.

Is this something new, or was it like this for a long time? From here it looks like server may be running out of capacity or being not reliable with uptime.

If anything, I have a dedicated server which is mostly idle (hosting some old pet projects that I can't bring myself to shutdown due to some remaining user activity).

It's characteristics are:

Intel Core i7-920 Quad Core
24 GB DDR3 RAM
2x 750 GB SATA II HDD

We could move OYP there if everyone and current admin is okay with it.

But if you pass on it, I will understand. After all, I am pretty new here, and moving OYP there is similar to granting me admin rights. Person with access to website database can basically do anything.

It happens quite a bit, like about every 2 months or so.

The software that powers the site isn't the most stable. Usually the admins have to reindex the DBs once in a while and restart the server and everything is good for a while.

If a migration would be possible, then we should strive for a software update too.
But moving this site is like moving an ice sculpture: it takes very little to crash this plane with no survivors.
Not saying we shouldn't try one day in the distant future, but just thinking about it makes me sweat profusely.

Well we did once a migration ages ago, it did go well without any hiccup in general, I think only user accounts was an issue there and mainly having to recreate them.

Shit....the memory is foggy but I am recalling the dread of suffering the original OYP Host....*shudders*

Thanks, now situation is more clear.

Let's mark it as "possible backup plan in case things make a turn for worse".

we have a dedicated host that has good uptime. Whenever the site goes down its usually because of posgresql crapping out and not anything to do with the hosting in general. (which we can't really fix).

The solution is to reboot the sql server and the site comes back up, but it does require manual intervention.

the first time we migrated it was fairly painless, with the exception of some problems with users forgetting passwords and us having to manually retrieve them. Upgrading to DB2 wasn't much different as far as I recall.

poweryoga said:

we have a dedicated host that has good uptime. Whenever the site goes down its usually because of posgresql crapping out and not anything to do with the hosting in general. (which we can't really fix).

The solution is to reboot the sql server and the site comes back up, but it does require manual intervention.

Maybe some sort of automated monitoring software like Nagios, which would execute same shell commands as you do manually?

poweryoga said:

we have a dedicated host that has good uptime. Whenever the site goes down its usually because of posgresql crapping out and not anything to do with the hosting in general. (which we can't really fix).

The solution is to reboot the sql server and the site comes back up, but it does require manual intervention.

the first time we migrated it was fairly painless, with the exception of some problems with users forgetting passwords and us having to manually retrieve them. Upgrading to DB2 wasn't much different as far as I recall.

Rather than a migration, what if we use this server as a mirror/backup while trying to update Danbooru and Postgres to a hopefully better version?
As was discussed in other threads, currently the system is running in debug mode and can't be switched to the correct mode.
A possible option is to make a backup on EasyDeibu's server, then update OYP's system to the current stable version of Danbooru. There's a chance doing this will also fix most database problems. If everything crashes and burns, we can simply take back the backed up version.
I'm aware that it's not as easy as it sounds, but we could try it at some point if EasyDeibu is still available.

I think we have something that automatically restarts the downed service, but that's Toawa's expertise and not mine. I tend to avoid logging into the server as much as possible since I know enough to royally fuck up things, but not enough to fix it if the royal fuck up happens.

Honestly, neither Toawa nor I are willing to put in the time/effort to upgrade the existing danbooru 2 engine. If anything, since the framework is so old, you should start with a fresh copy of the latest stable build of danbooru 2 and then we can see if we can write migration scripts + offload the image assets to the new db. I'm honestly not a huge fan of DB 2, but it is what it is.

Danbooru 2 is lightweight enough that you can probably set it up on a local vm and play with it as you please, on whatever linux distro.

The problems we have are:

1) Can't turn off debug mode (wtf)
2) Our posgresql keeps losing its index / crashing / not compatible with the latest danbooru2 engine. So stuff crash and burns.
3) Not a web developer, and I have no desire to learn JS or Ruby, so no new development on our pages / bug fixes.
4) Don't have the desire/will to go through another migration. I don't touch web-server stuff on my day to day, and I have no desire to learn it.

At this point in this site's life, I'm not looking for anything ambitious. Toawa did all the hardwork last time, I just pay the bills and I think we were both surprised the migration went so smoothly.

If someone wants to take over backend server maintenance or do dev work on the frontend that's fine (since nobody's actively maintaining the site), but if something breaks and we can't bring the site back up that'll be the end of it. I don't have any desires to rollback or restart from something new and honestly, this site has been up for almost a decade and it's pretty incredible for something that started as a meme.