We have recovered...sorta

[Barrister]

Ah, so we stepped back to May 2010.  How young & naive we all were back in those days...

May 2nd to be precise.

Since my son was born May 31st, things have changed a lot since then...

[The Larch]

Spain still won the world Cup, you're not taking that away.

[DontSayBanana]

I was thinking perhaps we should consider getting Languish hosted on some kind of cloud site. Let someone else deal with all the backup and hardware crap.

Bandwidth is an issue, since we don't prune the message table or the user list.  200 members searching an archive of 2+ years could get us expensive or canned pretty quickly.

[Berkut]

I was thinking perhaps we should consider getting Languish hosted on some kind of cloud site. Let someone else deal with all the backup and hardware crap.

Bandwidth is an issue, since we don't prune the message table or the user list.  200 members searching an archive of 2+ years could get us expensive or canned pretty quickly.

I dunno, seems kind of unlikely that Languish is actually significant compared to what they deal with regularly. Although I guess I don't really know...

[Baron von Schtinkenbutt]

I was thinking perhaps we should consider getting Languish hosted on some kind of cloud site. Let someone else deal with all the backup and hardware crap.

Bandwidth is an issue, since we don't prune the message table or the user list.  200 members searching an archive of 2+ years could get us expensive or canned pretty quickly.

Berkut's suggestion does have me looking at cloud options, but your point is the key consideration.  It goes beyond Languish, too, since I was using the server as a fileserver and a platform for side projects.

At any rate, I've been really lucky the past 6 or so years that I've had so few major problems.  I do need to get the server out of my lab and into a more robust environment.  So, my current plan is to colocate.  Unless I hear bad things about them, I intend to use Colo Unlimited, since I can get a 2U slot for $50/month and they're local.

As for the server itself, I found a 2U dual Mini-ITX chassis.  I intend to exploit the second slot by building a remote backup server to run next to the web server.  The backup server is going to run on the current web server's hardware (the Atom D510 board) with a pair of WD Green 2TB drives connected as a JBOD array.  The new web server is going to use an Atom D525 board with a mirrored pair of WD RE4s.  In addition to the mirror, I'm going to configure the web server to make regular image dumps to the backup server.  I'm also going to regularly pull critical data backups, like database dumps, to either Dropbox or one of my lab machines so I have backups at two sites (the backup server is primarily for giving me remote backups of my lab boxes).

The RE4s should stand up to the server load much better than the consumer-grade Seagate shit I've been using.  The Greens should be sufficient for a low-load backup server.  Getting both into a real datacenter will mean far fewer service interruptions.  It also means I can ditch this T1.

[Baron von Schtinkenbutt]

I was thinking perhaps we should consider getting Languish hosted on some kind of cloud site. Let someone else deal with all the backup and hardware crap.

Bandwidth is an issue, since we don't prune the message table or the user list.  200 members searching an archive of 2+ years could get us expensive or canned pretty quickly.

I dunno, seems kind of unlikely that Languish is actually significant compared to what they deal with regularly. Although I guess I don't really know...

With Amazon at least, what's going to kill us is not bandwidth, its IOPS.  The cost of EBS is heavily dependent on that.  I haven't researched other cloud options yet to see if there is a more favorable option.

[Baron von Schtinkenbutt]

So, I dug up and restored the May backup of the board.  The short version: I lost another fucking Seagate drive.  I'll discuss more after I get out of class.

Please report any and all issues and I'll try to get on them quickly.  I still have a lot of configuration to do and a shitload of real-life obligations at the moment.

I personally never had a problem with Seagate, but one of my friends, operating in a server environment (university) has lost 3 during the last year...  Maybe they ain't cut out for this kind of work.

I've only ever lost Seagate drives in server environments (except the one that was DOA), so I'm inclined to agree.

[Baron von Schtinkenbutt]

It isn't clear to me that a straight mirror would be the best option.  I say let's go ahead and put together the cash to do a proper RAID controller + drives setup.

Moldy, if you can get an approximate price tag, I (or someone else, if they'd rather) can take care of getting together the cash commitments.

A straight mirror as the only hedge against hardware failure, I agree.  With regular off-machine image backups though I don't think a full-on RAID is necessary.  Plus, LVM2 is pretty damn good.

[Admiral Yi]

   How much of that did you just make up Moldy?

[Darth Wagtaros]

It ain't ZFS though.

[The Brain]

That sounds about right, vM.

[Tamas]

Well I guess you could just have snapshot.exe do regular backups to another drive. I LOVE that little software. Well, no wonder, when you consider that the regular backup solution of my employer starts with a T...

[The Brain]

Timmay!

[Slargos]

Well I guess you could just have snapshot.exe do regular backups to another drive. I LOVE that little software. Well, no wonder, when you consider that the regular backup solution of my employer starts with a T...

..aking a picture of the serial number for the purpose of insurance fraud?

[Jacob]

Good to be back