How big is the drive Metafilter is located on? February 17, 2004 7:25 PM Subscribe
Matt, this is a question to you (as well as a push to the other users here).
How big is the drive Metafilter (or all of your stuff) is located on? Like many others, I lived in a dark world without MeFi and ask:
Have you thought about a raid level 1?
How big is the drive Metafilter (or all of your stuff) is located on? Like many others, I lived in a dark world without MeFi and ask:
Have you thought about a raid level 1?
Was drive failure the issue? I thought it was a problem with the cooling system that necessitated RAM replacement. In any event, a cooling system failure often damages drives, even if the damage isn't visible right away. Redundant disks are a good idea for sure, and an option I'm quite sure Matt is aware of.
posted by scarabic at 7:58 PM on February 17, 2004
posted by scarabic at 7:58 PM on February 17, 2004
They make servers with redundant fans and power supplies, RAID, robot backups, etc. I'm sure this is news to Matt. And I'm sure matt can cough up the 10K to get a badass server. Sure, why not? As long as we're whining about our pony, why not make it an expensive one. Hell, it's not our money.
posted by y6y6y6 at 8:49 PM on February 17, 2004
posted by y6y6y6 at 8:49 PM on February 17, 2004
If we can raise enought money to send the Mathowies to Iceland, I have a feeling we can cover the cost of a modest, yet reliable server.
posted by 4easypayments at 9:10 PM on February 17, 2004
posted by 4easypayments at 9:10 PM on February 17, 2004
I sort of helped Matt look into some possible upgrades. Raid is definitely something he is considering.
I'd love for MeFi to find a way to buy him a dual Opteron server with about 2GB RAM, and 400 or 500GB in Raid 0+1.
posted by riffola at 9:14 PM on February 17, 2004
I'd love for MeFi to find a way to buy him a dual Opteron server with about 2GB RAM, and 400 or 500GB in Raid 0+1.
posted by riffola at 9:14 PM on February 17, 2004
Your obviously missing the point, y6. You can produce an exceedingly reliable and powerful data system very cheaply now.
Look at it like this.
Promise Technology 4 port Serial ATA RAID PCI card, $139.99.
120GB Western Digital SATA hard drive with 8mb Cache, average seek time: 8.9ms, $119.99 x2
1 or 2 Panaflo 80mm fans, 5.99 to 7.99 apiece. $391.95 to 395.95 for a rock solid RAID 0 setup, no frills, high performance. These robots you talk about, do they vibrate?
posted by Keyser Soze at 9:16 PM on February 17, 2004
Look at it like this.
Promise Technology 4 port Serial ATA RAID PCI card, $139.99.
120GB Western Digital SATA hard drive with 8mb Cache, average seek time: 8.9ms, $119.99 x2
1 or 2 Panaflo 80mm fans, 5.99 to 7.99 apiece. $391.95 to 395.95 for a rock solid RAID 0 setup, no frills, high performance. These robots you talk about, do they vibrate?
posted by Keyser Soze at 9:16 PM on February 17, 2004
er I should clarify that 400 or 500GB is the end result of Raid 0+1, so each drive would have to be 200GB or 250GB. In fact I am not sure if Raid 0+1 will result in that much space using 200GB drives, but anyway 500GBish with RAID's speed and backup is what is needed.
posted by riffola at 9:17 PM on February 17, 2004
posted by riffola at 9:17 PM on February 17, 2004
(i meant RAID 1, not 0)
posted by Keyser Soze at 9:19 PM on February 17, 2004
posted by Keyser Soze at 9:19 PM on February 17, 2004
"If we can raise enought money to send the Mathowies to Iceland........"
It took Miguel burning the last vestiges of his credibility to raise enough for the iceland thing. Before he posted that mother of all bad posts we were way short. He really doesn't get enough thanks for that.
I figure it will only take 3-4 old timers tossed in a fire pit to raise enough for a killer server. Who's first?
posted by y6y6y6 at 9:22 PM on February 17, 2004
It took Miguel burning the last vestiges of his credibility to raise enough for the iceland thing. Before he posted that mother of all bad posts we were way short. He really doesn't get enough thanks for that.
I figure it will only take 3-4 old timers tossed in a fire pit to raise enough for a killer server. Who's first?
posted by y6y6y6 at 9:22 PM on February 17, 2004
send the Mathowies to Iceland
I just want to say that I think some brave Mefite should start a band called The Mathowies.
/derail
posted by anastasiav at 9:26 PM on February 17, 2004
I just want to say that I think some brave Mefite should start a band called The Mathowies.
/derail
posted by anastasiav at 9:26 PM on February 17, 2004
Isn't there an issue with the noise of extra fans, given that the server is in someone's wardrobe? Also, a more powerful server would mean more heat, therefore more fans, therefore more noise. Than again, maybe the current server could be auctioned off to pay for a new one? Who has plenty of money and would like to have Skippy II on their mantelpiece? *looks at Miguel*
posted by dg at 9:55 PM on February 17, 2004
posted by dg at 9:55 PM on February 17, 2004
I'm sure matt will have no problem trading in his Iceland trip for a new kickass server. Or hell, save the money by not having any children. He's a team player I'm tellin' ya! Dedication taken to a new level.
;)
posted by justgary at 10:12 PM on February 17, 2004
;)
posted by justgary at 10:12 PM on February 17, 2004
2 low noise fans would introduce a minute noise difference, especially since the odds of others fans in the system being louder are quite possible.
Even so, you could spend another 3 dollars per fan and buy Vantec stealth fans. 21 Decibels at 2100 RPM. (SF8025L)
posted by Keyser Soze at 10:15 PM on February 17, 2004
Even so, you could spend another 3 dollars per fan and buy Vantec stealth fans. 21 Decibels at 2100 RPM. (SF8025L)
posted by Keyser Soze at 10:15 PM on February 17, 2004
Dear Mathowie,
On behalf of 17,000 fellow users, I would humbly like to request that you refrain from having children, as, in the opinion of the electorate, offspring would increase the chances of bad posts going undeleted, as well as the chances of yoplait being spilled into the server. While we fully understand your natural paternalistic urges, we, the electorate, must unfortunately veto the decision for children. Understand that we share your pain at this decision fully, and will send you countless digital photographs of our own children, that you might live a brutal, surrogate existence through our own offspring.
Also, do not forget that according to last quarter's budget, you are required to purchase and install a new RAID device no later than March the First.
Thank you,
-Kaibutsu
posted by kaibutsu at 10:21 PM on February 17, 2004
On behalf of 17,000 fellow users, I would humbly like to request that you refrain from having children, as, in the opinion of the electorate, offspring would increase the chances of bad posts going undeleted, as well as the chances of yoplait being spilled into the server. While we fully understand your natural paternalistic urges, we, the electorate, must unfortunately veto the decision for children. Understand that we share your pain at this decision fully, and will send you countless digital photographs of our own children, that you might live a brutal, surrogate existence through our own offspring.
Also, do not forget that according to last quarter's budget, you are required to purchase and install a new RAID device no later than March the First.
Thank you,
-Kaibutsu
posted by kaibutsu at 10:21 PM on February 17, 2004
A raid donation request in comparison to the last donations would be quite easy to do, benefitting all of us.
posted by Keyser Soze at 10:27 PM on February 17, 2004
posted by Keyser Soze at 10:27 PM on February 17, 2004
I'm sure that Matt and his hosting partner(s) have discussed some options already, but I think that framing the problem in terms of getting a bigger, better machine might miss much of the point.
Simply having replacement parts on hand would have drastically shortened the downtime, from what I've been able to understand, and having some kind of backup/colo is not a bad idea either.
From what I've been able to glean from the status updates, the timing of someone's vacation had a multiple day impact as well. We need to come to grips with the kind of hosting arrangement we're running on here. It's not a 24/7 data center special, and probably couldn't afford to be, on the given budget, without sacrificing some necessary access/support/flexibility/whatever.
So, Keyser, a more robust platform would involve a lot of things besides simply a better machine or adding a RAID rig. Some of these issues you can throw money at, and some you need to put time and thought and work into as well.
posted by scarabic at 10:59 PM on February 17, 2004
Simply having replacement parts on hand would have drastically shortened the downtime, from what I've been able to understand, and having some kind of backup/colo is not a bad idea either.
From what I've been able to glean from the status updates, the timing of someone's vacation had a multiple day impact as well. We need to come to grips with the kind of hosting arrangement we're running on here. It's not a 24/7 data center special, and probably couldn't afford to be, on the given budget, without sacrificing some necessary access/support/flexibility/whatever.
So, Keyser, a more robust platform would involve a lot of things besides simply a better machine or adding a RAID rig. Some of these issues you can throw money at, and some you need to put time and thought and work into as well.
posted by scarabic at 10:59 PM on February 17, 2004
Well, fans won't help if the server is in an unventilated closet without dedicated AC. You'd just be blowing hot air back into the box. Clearly we need to raise enough for a dedicated machine room. 100K should be enough.
posted by y6y6y6 at 12:03 AM on February 18, 2004
posted by y6y6y6 at 12:03 AM on February 18, 2004
100K should be enough.
Heh.
I promise you that if we send matt 100 k the next day metafilter/wholelottanothing/tenyears/haughey.com/tivo etc will have vanished from existence and the missing persons division of the fbi will not be able to find him under his 'new' name.
At least, if I were matt.
posted by justgary at 12:26 AM on February 18, 2004
Heh.
I promise you that if we send matt 100 k the next day metafilter/wholelottanothing/tenyears/haughey.com/tivo etc will have vanished from existence and the missing persons division of the fbi will not be able to find him under his 'new' name.
At least, if I were matt.
posted by justgary at 12:26 AM on February 18, 2004
I think a broken fan was nothing compared to what could happen if the hard drive crashed. I don't want to come off as needing this website to live, but I like the thought of data redundancy.
posted by Keyser Soze at 2:42 AM on February 18, 2004
posted by Keyser Soze at 2:42 AM on February 18, 2004
Why not just hire a full-time server tech to constantly monitor the web server? I'll bet there's room in that closet for him to crouch in there.
Riff -- what are you doing for the next 30 years?
posted by crunchland at 2:43 AM on February 18, 2004
Riff -- what are you doing for the next 30 years?
posted by crunchland at 2:43 AM on February 18, 2004
It seems like there are two ideas going on at once, one side that thinks a small investment can double the data redundancy, and another group that is comparing a sub 500 dollar investment to making a mefite member sit in the server closet and costing Matt Haughey thousands of dollars.
Understand that no matter what you do, there is no 100 percent guarantee on data redundancy, no matter what steps you take.
There are 17 thousand members of a website that has had books written on it, has been mentioned in magazine articles, carrying the culmination of information and ideologies for the last 5 years (as a website) that has very recently been offline for two weeks, and I feel a community effort to prove we may still read our posts another 5 years from now for a few bucks from a few members is worth the effort.
There are many of you who may disagree with me, or cite other ways that we can all lose a piece of the internet, but I feel using RAID to backup our data can be the most cost effective and physically effective way to invest our time and effort.
posted by Keyser Soze at 3:10 AM on February 18, 2004
Understand that no matter what you do, there is no 100 percent guarantee on data redundancy, no matter what steps you take.
There are 17 thousand members of a website that has had books written on it, has been mentioned in magazine articles, carrying the culmination of information and ideologies for the last 5 years (as a website) that has very recently been offline for two weeks, and I feel a community effort to prove we may still read our posts another 5 years from now for a few bucks from a few members is worth the effort.
There are many of you who may disagree with me, or cite other ways that we can all lose a piece of the internet, but I feel using RAID to backup our data can be the most cost effective and physically effective way to invest our time and effort.
posted by Keyser Soze at 3:10 AM on February 18, 2004
trouble is, it's not your time and effort. reconfiguring hardware is a pain in the backside - it always takes more effort and time than you bargained for and always introduces more problems down the line. and matt has already said he's no great sysadmin. what's more, as far as i understand things, he's not physically near the machine, so one person is going to be messing around with hardware and then he's going to have to try and get everything running remotely. i don't blame him for not wanting to be bothered.
and why the emphasis on disk reliability anyway? as long as the code, user data and general config are backed up somewhere, what's the big deal with losing old threads? they're not exactly great literature - it's the community that makes this place, not the data mining.
raid would not have prevented the outage we just had.
posted by andrew cooke at 4:01 AM on February 18, 2004
and why the emphasis on disk reliability anyway? as long as the code, user data and general config are backed up somewhere, what's the big deal with losing old threads? they're not exactly great literature - it's the community that makes this place, not the data mining.
raid would not have prevented the outage we just had.
posted by andrew cooke at 4:01 AM on February 18, 2004
Losing old threads in AskMe would be a big deal -- I was going nuts during the downtime trying to find by other means the answer to one of my questions. If it had been up we'd have been fine.
posted by bonaldi at 4:12 AM on February 18, 2004
posted by bonaldi at 4:12 AM on February 18, 2004
I would humbly like to request that you refrain from having children...
*snip!*
posted by quonsar at 5:13 AM on February 18, 2004
*snip!*
posted by quonsar at 5:13 AM on February 18, 2004
There are 17 thousand members of a website that has had books written on it, has been mentioned in magazine articles, carrying the culmination of information and ideologies for the last 5 years (as a website) that has very recently been offline for two weeks...
I think some of you are under the mistaken impression that all of this really matters.
posted by crunchland at 5:46 AM on February 18, 2004
I think some of you are under the mistaken impression that all of this really matters.
posted by crunchland at 5:46 AM on February 18, 2004
Maybe some members with sufficient disk space and bandwidth could volunteer to host static mirrors. This would allow the archives to remain up during server downtime or connection outages.
posted by Songdog at 6:31 AM on February 18, 2004
posted by Songdog at 6:31 AM on February 18, 2004
I would humbly like to request that you refrain from having children...
I was just talking about how babies ruin the internet.
posted by jennyb at 7:07 AM on February 18, 2004
I was just talking about how babies ruin the internet.
posted by jennyb at 7:07 AM on February 18, 2004
I have to agree - old threads are well worth saving. There are certain things that I head straight to the mefi archive to find out. I know I'm not the only one to find old threads damn useful.
posted by Skaramoosh at 7:16 AM on February 18, 2004
posted by Skaramoosh at 7:16 AM on February 18, 2004
"Maybe some members with sufficient disk space and bandwidth could volunteer to host static mirrors."
I've thought about this. But there would be a problem with syncing things up. And also the issue of many people having read/write access to the user tables and the ability to delete things they don't like.
I have plenty of bandwidth right now, but do you want me deciding which posts get the ax? I certainly hope not.
posted by y6y6y6 at 7:16 AM on February 18, 2004
I've thought about this. But there would be a problem with syncing things up. And also the issue of many people having read/write access to the user tables and the ability to delete things they don't like.
I have plenty of bandwidth right now, but do you want me deciding which posts get the ax? I certainly hope not.
posted by y6y6y6 at 7:16 AM on February 18, 2004
What about just collecting enough for a reliable tape backup system? I'd kick in some money for that. It wouldn't require reconfiguration of the hardware, just slapping an external SCSI tape drive on.
posted by mrbill at 7:32 AM on February 18, 2004
posted by mrbill at 7:32 AM on February 18, 2004
I like how everyone just assumes there aren't regular backups already. Does anyone here have a site they feel is important enough to have maintained for four years and then not backed it up regularly?
posted by y6y6y6 at 7:47 AM on February 18, 2004
posted by y6y6y6 at 7:47 AM on February 18, 2004
a website that has had books written on it
I know Miguel's written enough here for a book, but has he had it published?
*goes off to google Metafiltro: um caso de amor sem fim*
posted by languagehat at 8:38 AM on February 18, 2004
I know Miguel's written enough here for a book, but has he had it published?
*goes off to google Metafiltro: um caso de amor sem fim*
posted by languagehat at 8:38 AM on February 18, 2004
if each of us talked to the IT departments (or equivalents) at the places we worked, perhaps we could cobble together a low-tech solution? a fan here, a hard drive there, a CPU or two for clustering. perhaps if someone could come up with a shopping list, the rest of us can start talking to our peoples. can we simply throw old hardware at this problem?
posted by VulcanMike at 9:10 AM on February 18, 2004
posted by VulcanMike at 9:10 AM on February 18, 2004
y6y6y6, I wasn't proposing a full mirror of the data, just a snapshot of the generated pages, a static archive. If several sites were to host mirrors then tampering would be detectable unless it was coordinated at all of the mirrors. If people are really concerned about this then an authentication system could be set up (using MD5 digests or something similar) but I would think that Matt's trust of the mirror admins would be sufficient for most of us.
posted by Songdog at 12:15 PM on February 18, 2004
posted by Songdog at 12:15 PM on February 18, 2004
While we all plot methods to keep our addiction alive it might be nice to join this list.
Matt has in place a method where we can financially support him via Amazon or PayPal. Kudo's to those of you already on the list.
With an active donation campaign, (Matt really ought to put it on the FP/SideBlog or maybe someone FPP it), Matt might be able to afford the resources necessary to make the dark days of last week just a distant memory.
posted by geekyguy at 1:01 PM on February 18, 2004
Matt has in place a method where we can financially support him via Amazon or PayPal. Kudo's to those of you already on the list.
With an active donation campaign, (Matt really ought to put it on the FP/SideBlog or maybe someone FPP it), Matt might be able to afford the resources necessary to make the dark days of last week just a distant memory.
posted by geekyguy at 1:01 PM on February 18, 2004
It took Miguel burning the last vestiges of his credibility to raise enough for the iceland thing.
And I thought it was *my* $20 that put us over the top.
I'd pitch in again for whatever solution theborg membership wants to purchase for our fearless leader.
posted by me3dia at 1:03 PM on February 18, 2004
And I thought it was *my* $20 that put us over the top.
I'd pitch in again for whatever solution the
posted by me3dia at 1:03 PM on February 18, 2004
Matt is the single failure mode. We must clone him.
posted by substrate at 2:16 PM on February 18, 2004
posted by substrate at 2:16 PM on February 18, 2004
Jesus, an enterprising drug dealer would make a total fortune off of some of you junkies.
Why don't we just crucify Haughey, and get it over with?
posted by crunchland at 3:24 PM on February 18, 2004
Why don't we just crucify Haughey, and get it over with?
posted by crunchland at 3:24 PM on February 18, 2004
Naaa... let it crash once in a while to create some popularity. currently Meta-Filter is the #1 search on Daypop Top 40 Links.
posted by thomcatspike at 4:05 PM on February 18, 2004
posted by thomcatspike at 4:05 PM on February 18, 2004
I have plenty of bandwidth right now, but do you want me deciding which posts get the ax? I certainly hope not.
posted by y6y6y6 at 7:16 AM PST on February 18
You might be an old timer pal, but you've f**k all recovery.
posted by sgt.serenity at 5:37 PM on February 18, 2004
posted by y6y6y6 at 7:16 AM PST on February 18
You might be an old timer pal, but you've f**k all recovery.
posted by sgt.serenity at 5:37 PM on February 18, 2004
Just to clarify a little bit:
The fan that died was the fan that sits directly atop the heatsink on one of the Athlon processors. Doesn't matter how warm or cool it is in the case at that point. There wasn't any other hardware hurt in the process, though; all mentions of memory replacement are because of lingering issues with the motherboard not wanting to reboot on occasion that seems to be related to memory self-tests.
The server does, indeed, sit in a closet here, but an open-faced closet that has three other computers in it. Not a ton of heat buildup, but some. (Alas, it's a piddly little pre-war Brookline, MA closet, so anyone volunteering to live inside it to be the constant server monitor better be tiny!)
That is all.
posted by delfuego at 7:04 PM on February 18, 2004
The fan that died was the fan that sits directly atop the heatsink on one of the Athlon processors. Doesn't matter how warm or cool it is in the case at that point. There wasn't any other hardware hurt in the process, though; all mentions of memory replacement are because of lingering issues with the motherboard not wanting to reboot on occasion that seems to be related to memory self-tests.
The server does, indeed, sit in a closet here, but an open-faced closet that has three other computers in it. Not a ton of heat buildup, but some. (Alas, it's a piddly little pre-war Brookline, MA closet, so anyone volunteering to live inside it to be the constant server monitor better be tiny!)
That is all.
posted by delfuego at 7:04 PM on February 18, 2004
I think the idea that the server lives in some guys closet and may go down at any time for an unknown duration only adds to the coolness here.
posted by y6y6y6 at 7:20 PM on February 19, 2004
posted by y6y6y6 at 7:20 PM on February 19, 2004
As it went down last night and again for most of today, just to remind us to bow down each and every night in the direction of Brookline, thanking our lucky stars for the days when MeFi is up and running. Whew, we're back again! Thanks to Jason and Matt for whatever you had to do to get it up and running again.
posted by Lynsey at 7:36 PM on February 19, 2004
posted by Lynsey at 7:36 PM on February 19, 2004
I feel that it is inappropriate of us to discuss Matt's "hosting partner." What he does on his own time is his business, not ours.
posted by mecran01 at 8:23 PM on February 19, 2004
posted by mecran01 at 8:23 PM on February 19, 2004
Shouldn't posts about RAID be filed in the "Bugs" category rather than "Metafilter Related"?
Ha ha ha! Because of the bug spray called Raid! Ha ha! Ha.. ha ha ha... ha...a..ha... ha. Lord, I'm so tired of living like this.
posted by Hildago at 8:37 PM on February 19, 2004
Ha ha ha! Because of the bug spray called Raid! Ha ha! Ha.. ha ha ha... ha...a..ha... ha. Lord, I'm so tired of living like this.
posted by Hildago at 8:37 PM on February 19, 2004
Easy there, Hildago.
Hello 911? I've got someone here suffering from late stage Algernon Syndrome. Yes, he's gone into nyuk-nyuk shock. Oh no - he knows he's not funny, but he doesn't care anymore. Yes, I've slapped him - and he poked me in the eyes for my trouble. Right now? He's humming "Three Blind Mice" and eating a pillow. No, I'm sorry - I'm sure a warm bath and a massage would help him - but I've got a dinner engagement. Could you send someone right over? His name is Hildago. He's dressed like a rubber chicken, he's wearing a squirting flower, he's sitting on a whoopee cushion, and he's throwing meringue pies at himself. Yes, it is a shame. He used to be so courtly, you know.
posted by Opus Dark at 12:16 AM on February 20, 2004
Hello 911? I've got someone here suffering from late stage Algernon Syndrome. Yes, he's gone into nyuk-nyuk shock. Oh no - he knows he's not funny, but he doesn't care anymore. Yes, I've slapped him - and he poked me in the eyes for my trouble. Right now? He's humming "Three Blind Mice" and eating a pillow. No, I'm sorry - I'm sure a warm bath and a massage would help him - but I've got a dinner engagement. Could you send someone right over? His name is Hildago. He's dressed like a rubber chicken, he's wearing a squirting flower, he's sitting on a whoopee cushion, and he's throwing meringue pies at himself. Yes, it is a shame. He used to be so courtly, you know.
posted by Opus Dark at 12:16 AM on February 20, 2004
You are not logged in, either login or create an account to post comments
posted by Keyser Soze at 7:38 PM on February 17, 2004