[Keiro]

Hi, everyone!

I am planning to migrate DS to the new server that has multiple network providers, in the next few days. I will try to make sure that it's not during the day time. Perhaps late in the night, when there is few users online, so the switch goes off smoothly. I will be leaving this notice up for 24 hours and do the transition late tomorrow night.

What does this mean for us? It means that no matter how a network outage happens, there will always be a backup provider, rerouting connections through the backup connections while the main one is fixed.

This means that the server will be going in this stage, as Kradorex described:

Kradorex Xeron, on 23 March 2010 - 04:51 AM, said:

It's a stopgap measure - Since it would be days and not merely hours to move the "official" main server to another physical facility, the current server and the backup server will be sync'd together as closely as possible, if not exact for a period of time. From there, the actual DNS (domain name) will be pointed at the backup server instead of the main server, then the current server will be taken offline and the forums will be operational from the backup server, at which time the main server will be physically re-located (it's a ShatteredTears/ EidolonHost-owned machine) to another facility with more reliable and redundant network connectivity. Once this relocation is completed, a sync will occur again, this time, from the backup server to the main server, as so that no content is lost.

During the synchronization time periods, the forums will be locked as to prevent posts from being made (ergo lost) during transitions between servers.

For those who prefer point form, the schedule will look a lot like this (without timestamps, no calculated plans have been made yet):

- Announcement made
- Forums de-activated on main server
- Database backup made and transferred to backup server
- Forum software copied and transferred to backup server
- DNS (domain name) re-directed to backup server, Forums re-activated on backup server
- Main server taken offline and shipped to new facility
Upon arrival of main server to new facility, basically the same would happen, only swap the servers.

Downtime expected: 4 hours maximum, 1 hour minimum for each cycle.

Expected downtime? 4 hours to 6 hours, maximum. One hour maximum for each cycle, give or take an extra 30 minutes, for security's sake.

The new server is online right now, but I'm securing it at the moment. In other words, I'm making sure that the risk that DS is hacked is as low as possible. (Preferably 0% but that's a dream that'll never be realized...)

At this time, the servers will synchronize data between the two until it's time to make the switch.

[dBm]

Let's call it a payback for not having the site die during the last Comiket~

[zjhentohlauedy]

during the night for you guys?

that means morning here huh...

could you do it over the week. so it will match with the time i'll be gone XD

[Zephiel]

dBm, on 29 March 2010 - 07:41 PM, said:

Let's call it a payback for not having the site die during the last Comiket~

=w=b

[Ninji]

Ninji of the night here. Oh well I don't mind.

[Purianite]

I'm sure I'll be asleep around that time anyhow. Plus, it's definitely worth the time.

[Stalker-san]

Night for whom?

[Maiorem]

Stalker-san, on 30 March 2010 - 02:09 PM, said:

Night for whom?

Good question.

[Karusune Kiyoraka]

I would assume night for the US (not sure about timezone) considering that Keiro is in the US. It would also be night for Canada, which is where Kradorex is as well.

[Keiro]

Karusune Kiyoraka, on 30 March 2010 - 01:34 AM, said:

I would assume night for the US (not sure about timezone) considering that Keiro is in the US. It would also be night for Canada, which is where Kradorex is as well.

Correctamundo.

[Keiro]

dBm, on 29 March 2010 - 08:41 PM, said:

Let's call it a payback for not having the site die during the last Comiket~

Eh?

[GhostNappa]

hmm, so down during the morning for me too. Oh well. I will lay in .

[Keiro]

We've begun the process of synchronizing the servers.

We are monitoring the status of the synchronization at this time.

The server will NOT go down at this time.

[Keiro]

>_>

Well, it did. That was us doing the transfers.

At least everything seems to have made it over correctly. Do please let me, Karusune, Kradorex and Holy know, though, if there's any missing posts.

What happened, was that the automatic transfers had failed. Why? Because MySQL choked and died on DS's HUGE SQL file. (About 1.5GB in size, if I remember correctly.) This screwed up a lot of things and took time to fix. Once that was done, we did it the slow, hard way. Transferring every last bit, manually and forcing MySQL to eat the SQL file.

And the end.

Thank you, and I apologize for the extended downtime during the transfer to the new Datacenter.

Provided that the server does not have a hardware failure, with several network providers in place, it should provide us with 99.9% uptime reliability, as whenever any one network provider fails, all routes will automatically switch over to the next working network provider.

Which is what we need, considering that DS gets a LOT of attention.

At this time, both servers will stay online, and anything missing will be retrieved from the original server.

[Full Throttle]

Keiro, on 31 March 2010 - 04:16 PM, said:

What happened, was that the automatic transfers had failed. Why? Because MySQL choked and died on DS's HUGE SQL file. (About 1.5GB in size, if I remember correctly.) This screwed up a lot of things and took time to fix. Once that was done, we did it the slow, hard way. Transferring every last bit, manually and forcing MySQL to eat the SQL file.

I lol'd. XD

[Keiro]

Full Throttle, on 31 March 2010 - 07:42 PM, said:

I lol'd. XD

>_> I was actually somewhat expecting this to happen, as MySQL does have a limit as to how much it can take all at once through the WHM interface.

[pegasoltaeclair]

No matter.

If you guys felt that the synch was necessary, then I'm sure it'll help.
Besides, the downtime was good for some people. Stopped me from randomly browsing DS while I was supposed to be doing homework for a bit.

[Full Throttle]

Keiro, on 31 March 2010 - 10:24 PM, said:

>_> I was actually somewhat expecting this to happen, as MySQL does have a limit as to how much it can take all at once through the WHM interface.

ya but I assumed that limit would be higher than a few gigs. ._.

[Keiro]

Full Throttle, on 01 April 2010 - 04:33 PM, said:

ya but I assumed that limit would be higher than a few gigs. ._.

Through the commandline, yeah.

Not through WHM/cPanel, which is how I'd initially started the synchronization. It chokes and dies if done through WHM/cPanel for some reason.

[Full Throttle]

Keiro, on 02 April 2010 - 09:53 AM, said:

Through the commandline, yeah.

Not through WHM/cPanel, which is how I'd initially started the synchronization. It chokes and dies if done through WHM/cPanel for some reason.

that explains it.