• Fur Affinity Forums are governed by Fur Affinity's Rules and Policies. Links and additional information can be accessed in the Site Information Forum.

Site down again?

Jabby1010

New Member
UPDATE TO THE UPDATE:

Okay, not quite accurate. It seems they had to do some swapping/moving/rearranging in general, which required unplugging and re-plugging some stuff, and just got the NIC cables swapped by accident.

As net-cat said in the FA Status LiveJournal, "angry letters have been sent."


Humm, and you pay these people to host this server...GOOD CHOICE! :rolleyes:
 

foxystallion

Born Furry
Because it didn't need to be.

Problem is looking to be caused by power fluctuations at the colo, which is not something we can control (except by switching colos again, which I think was on the to-do list anyway).

The sensor data suggests that the power fluctuations were of fairly short duration. Perhaps a few hundred dollars spent on an uninterruptable power supply with a battery bank adequate for 20 to 60 minutes would be a worthwhile investment. I live way out in the middle of nowhere in central Nevada with over 200 miles of transmission line between me and the nearest power plant, so I get a LOT of brief power dropouts. I have APC UPSs on all my computer equipment, and the investment has been very worthwhile. Hopefully, there is enough donation money left over to do this.
 

foxystallion

Born Furry
Dunno. AFAIK, there are not any solid plans for doing that, just speculation.

Please consider the alternative of a UPS. My APCs (with the highest joule rating surge suppressors I could get between them and the nasty power line) have done the job for years. Lightning has hit the distribution lines powering my home three times within 100 yards of me, and my computers have kept on running without a glitch. I must admit that I rewired my home so that it now has a single point ground system thereby avoiding ground loops that pick up EMP from lightning (I'm an aerospace weapons physicist), but I can't imagine colos not having single point grounds.
 

BillyRabbit

New Member
I'm so tired of everyone immediately jumping on everyone who gets mad that the site is down. I go to many different sites on the internet, and they seem to almost never go down. Granted, FurAffinity is a graphics message board--and one that has pretty heavy traffic. But, it still seems to go down for a few days or a week every three weeks or so.

The reason people get angry is because they donate to the site and the site. The site has to make money on ads, doesn't it?!? I know it's not a lot compared to being Fox and owning MySpace, but, it's still money.

GameFAQs used to only make money on adverts. It was pretty much coded and run by a guy and a very small team of friends. He got it big enough and sold it to CNet (owned by Ziff Davis, I believe?), and has since finally washed his hands of what was probably a nightmare for him to run.

But, guess what? That site very rarely was down. Granted, it was mostly text and all, but it still was rarely down.

The real problem is that FA is cool. Just like DA, you can watch people, talk to other users, and you can have your own stuff posted to your own sites and directories. That's much better than the *chan-style boards or the old VCL sorts.

It's just frustrating to have a cool spot on the net that routinely goes down and has all these problems, so.... Don't get immediately angry at everyone for "baw"-ing. I swear, jumping on people for complaining is the new "tl;dr." (waits for someone to see this post and write "tl;dr")
 

foxystallion

Born Furry
UPDATE:

We think we've identified the problem. Seems like someone at the colo decided it was a good idea to swap the cables on Trogdor's two NIC cards.

Aaaaarrrrrrrggggghhhhhhhh!!!! If only there were an uninterruptable intelligence supply...
 
Last edited:

Rhainor

Rawr.
Please consider the alternative of a UPS. My APCs (with the highest joule rating surge suppressors I could get between them and the nasty power line) have done the job for years. Lightning has hit the distribution lines powering my home three times within 100 yards of me, and my computers have kept on running without a glitch. I must admit that I rewired my home so that it now has a single point ground system thereby avoiding ground loops that pick up EMP from lightning (I'm an aerospace weapons physicist), but I can't imagine colos not having single point grounds.
This was mentioned in IRC. It is the colo's responsibility to provide us with stable, reliable power.

That said, the supposed "power fluctuations" in this case look to have been simply the colo guys unplugging/re-plugging the current UPS or whatever for maintenance/rearranging/whatever the hell they were doing that left us with swapped network cables, and not actual problems with the power supply.
 

Nidonemo

Member
I'd be a bigger fan of DA if they hadn't moved my work around without my permission. I heard rumor that it's because they want to keep up an image?

It's an art portal, what kind of image do you need?

In any case, who goes where when it hits the fan?
 
Somebody smack the guy who mixed up the cables. He made me lose a bet.
 

rednec0

Member
has the colo ever heard of calling owners and saying "hey we're doing this, this, and this so you might see something happen"? tell dragoneer and company to stand over them with a BIG stick in case they try pulling any funny business again. its stuff like this that make me lose more faith in the human race day-by-day.
 

Nidonemo

Member
...but we don't know what work life is for them over there. The "colo" or over at the "colo"...it could be very hectic and busy and this was an accident that they are very sorry about. I bet there's thousands of cables and keeping track of two is like looking at a plate of noodles and trying to follow two individual strands at once.
 
...but we don't know what work life is for them over there. The "colo" or over at the "colo"...it could be very hectic and busy and this was an accident that they are very sorry about. I bet there's thousands of cables and keeping track of two is like looking at a plate of noodles and trying to follow two individual strands at once.

As a teacher, I can be very sorry that in the hectic grading period I make a mathematical error causing a student to fail my course.

It might have been an honest mistake, but the student still has the right to be seriously cheesed off and demand it get fixed immediately.
 

fx1

Member
I was just thinking, pure speculation.. The new server has a lot of memory, which means it probably has a large cache and backlog of data "to-be-written" to the disk. Larger than the old server. Which means that sudden power outages are even more harmful, if the server doesn't commit the data to the disk often enough. I hope the database is consistent and also not rolled back too much. I wouldn't be surprised if the latest submissions before the outage went to black hole..
 

icehawk

Member
Please consider the alternative of a UPS. My APCs (with the highest joule rating surge suppressors I could get between them and the nasty power line) have done the job for years. Lightning has hit the distribution lines powering my home three times within 100 yards of me, and my computers have kept on running without a glitch. I must admit that I rewired my home so that it now has a single point ground system thereby avoiding ground loops that pick up EMP from lightning (I'm an aerospace weapons physicist), but I can't imagine colos not having single point grounds.


As counterintuitive as it sounds, you don't put machines in a datacenter on individual UPSes, since the EPO button has to kill all of the power in a facility.


UPDATE:

We think we've identified the problem. Seems like someone at the colo decided it was a good idea to swap the cables on Trogdor's two NIC cards.

This is totally why I label ports and both ends of a cable.
 

Nidonemo

Member
As a teacher, I can be very sorry that in the hectic grading period I make a mathematical error causing a student to fail my course.

It might have been an honest mistake, but the student still has the right to be seriously cheesed off and demand it get fixed immediately.

This is true, but would you rather the student be civil and ask politely, or would you have him jumping up and down swearing his head off?
 

AlexInsane

I does what I says on the box.
Having a bit of a site hiccup again?

I keep getting a white screen now.
 

yak

Site Developer
Administrator
Yes, there appears to be some TCP/IP issues again.
 

Ainoko

The Wild And Crazy Hybrid
Yes, there appears to be some TCP/IP issues again.


Thanks for the info! That helps to explain me getting into other member's userpages

members whose pages I am accessing right now,

1. inarolylan
2. chaos31
3.
 
Last edited:

yak

Site Developer
Administrator
*smiles* It doesn't matter who'se pages you are trying to access, it's a global problem. Basically, running out of network packet buffers.

I have just decreased the size of the socket listen queue and reduced the keepalive setting for HTTP connections. It should help offset the problem somewhat until I can look at this properly tomorrow.
 

yak

Site Developer
Administrator
Seems that the problem has temporarily went away. Number of ESTABLISHED connections decreased from 12k something to 7k something.
 

dmfalk

Member
Yes, OS X is basically Free BSD UNIX with a proprietary user friendly shell and various other doodads. For those who still doubt, just watch the screen while installing OS X and you will see Free BSD being installed along with the doodads.

Actually, it's BSD, not FreeBSD. It's fully licensed.

d.m.f.
 
Top