Slashdot Mirror
Welcome To The New Slashdot Server
After much blood, sweat and tears, the new server appears to be up.
It'll probably be a few hours yet while the DNS trickles over. We'll have a more extensive report describing the new hardware in the next couple of days... but first, we gotta iron out any kinks that pop up.
There's something not quite right about viewing Slashdot on a fast server ... sort of like listening to old Beatles LPs on mp3, or seeing the Blair Witch Project in THX.
Power corrupts. PowerPoint corrupts absolutely.
Search the archive for stories in the "Slashdot.org" category... You'll find quite a bit of slashdot ancient history. :)
--
All Glory To The Hypnotoad!
Just out of curiousity. What hardware was the old slashdot running under?
And, what were the biggest reasons for upgrading?
-Needed Faster CPU?
-Needed more memory?
-Needed a bigger network pipe?
-Needed clustering?
-___ ?
No! This server is MINE! Get off.
Posted by BSD-Pat:
We had an article on the beta a while back but:
arrowpoint CS series load balancer...
100 mbit pipe
cisco 6509/2 MSFC's
4 pIII650 web servers
2 pIII650 image servers
1 Dual pIII650 slashd
1 Quad xeon 550 DB server
the OS's are Linux 2.2.14 (Debian and RedHat)
the servers I think are apache 1.3.12+mod_perl
Umm, you might be checking the _old_ slashdot on Netcraft. I got these results.
64.28.67.48 is running Apache/1.3.12 (Unix) mod_perl/1.22 on Linux
The right setup for this would be something more like two Sun 420s serving pages and a 3500 for the DB/fileserver. And yes, they should be running Linux (with Solaris you'd need twice as many machines to compensate for the built-in Molasses [tm] feature).
may I sleep now?
I've had this problem for quite some time... haven't you noticed how we're continually reminded about Jon Katz's "recent" Hellmouth articles?
Way to go, now I can reload slashdot even faster during the day. :)
-- dieman - Scott Dier
*Joke*
I would like to be the first to claim that Slashdot began to go bad when they went to this new server.
There is plenty of room to play with the dietary suplements for the hamsters. However, within 18 months it will be necessary to consider upgrading to trained rats. Then again microsoft.com gets suprising performance using large quantities of bugs. There is also a rumor that Big Blue has an interesting multiplying server project under development using rabbits.
134340: I am not a number. I am a free planet!
Good Job!
/. off of that shared Alpha onto it's own system!
This feels like as much of a speed increase as when you moved
--
Holy speed batman! This is how slashdot always came up in my dreams, so fast I haven't even realeased the enter key and it's up :)
And stop looking at me funny, you dream about web surfing too.....don't you?
Finkployd
you'll note that the IP his link looks up at netcraft is 207.46.130.149, which, although it has no reverse-DNS, is served from an ip for which dns4.cp.msft.net is athoritive.
:-) nice try though. On a hunch, I forward-looked up www.microsoft.com (hey, trolls aren't usually that creative) and it's one of the 4 servers in that DNS round-robin.
that's not slashdot
The Matrix is going down for reboot now! Stopping reality: OK. The system is halted.
First of all, you need to do some research about what Apache does and doesn't cache.
Second, mod_mmap doesn't put anything in the HEAD request, so if they were to take this simple and adequate route to caching, you wouldn't have a clue about it. mod_mmap is totally transparent to you, the user. It's not totally transparent to the sysadmins (one of it's weaknesses, IMHO) but then if you're only serving up 60 images, that shouldn't be a big deal.
Finally though, I really think people make too big a deal about serving images off a separate server. Let's look at the bigger picture: Slashdot serves around a million page views a day. That's averaging out throughout the day at a whopping 11 requests/sec on page views. Now given that we've seen benchmarks of apache serving static content at 2000+ req/sec, I think it's probably not even breaking a sweat serving those images.
What slashdot really needs to do (and I know pudge knows this too) is get away from Apache::Registry. That and do some serious code cleanups, and maybe even try this app on Oracle. That's where you'll see real differences, IMHO.
Anyway, I suppose they say every little helps, so how about it Pudge? Is mod_mmap doing it's thang in there?
Matt. Want XML + Apache + Stylesheets? Get AxKit.
I read somewhere a trick to reduce the load on your Apache:
Set the caducity (word? I mean the time after it is no longer valid) to a date in the future (say after 1 month) for constant images.
Thus, people will cache (locally o proxily) the images and they won't request them from your browser, and images are a lot of bytes per file.
Of course, this is for constant images (formatting pixels, topics,...) not for banner ads, counters and doubleclick bigbrothers.
__
__
Men with no respect for life must never be allowed to control the ultimate instruments of death.
GW Bu
The good news is that response is definitely faster. The bad news (perhaps) is that this may be due to my being online at 5:50 EDT, when all the good little geeks who don't have to go to the airport this morning are still in bed. I'll try it later this morning, after the country wakes up.
/.-type site it's probably the best option for you. Good job.
However, loading a large text page (like this one) is noticeably faster, as are the image loads. And the preview is really fast. I remember when you were talking about this at Geek Pride last month - Exodus is a good shop and for a
- -Josh Turiel
-- Josh Turiel
"2. Do not eat iPod Shuffle."
[NRAdude@gtmi686dpp NRAdude]$ ping slashdot.org
PING slashdot.org (64.28.67.48): 56 data bytes
64 bytes from 64.28.67.48: icmp_seq=0 ttl=110 time=249.6 ms
64 bytes from 64.28.67.48: icmp_seq=1 ttl=110 time=229.9 ms
64 bytes from 64.28.67.48: icmp_seq=2 ttl=110 time=229.9 ms
64 bytes from 64.28.67.48: icmp_seq=3 ttl=110 time=219.9 ms
64 bytes from 64.28.67.48: icmp_seq=4 ttl=110 time=219.9 ms
64 bytes from 64.28.67.48: icmp_seq=5 ttl=110 time=210.0 ms
64 bytes from 64.28.67.48: icmp_seq=6 ttl=110 time=229.9 ms
--- slashdot.org ping statistics ---
7 packets transmitted, 7 packets received, 0% packet loss
round-trip min/avg/max = 210.0/227.0/249.6 ms
[NRAdude@gtmi686dpp NRAdude]$ queso slashdot.org
dude! wow! fast and efficient! Nosebleed! They're using Linux kernel 2.2.x
[NRAdude@gtmi686dpp NRAdude]$ logout
(today, at night, about 3:00am, can't get on slashdot.org, GRRRRRRR)
[NRAdude@gtmi686dpp NRAdude]$ ping slashdot.org
PING slashdot.org (64.28.67.48): 56 data bytes
64 bytes from 64.28.67.48: icmp_seq=6 ttl=110 time=666,666.666ms
--- slashdot.org ping statistics ---
666 packets transmitted, 1.666 packets received, 666% packet loss
round-trip min/avg/max = 6666.66/6666.66/6666.66 ms
[NRAdude@gtmi686dpp NRAdude]$ queso slashdot.org
eeeek! They're using Windows 2000 pseudo smp mode. leave in the name of Jesus! SIG15. Core dumped.
...
...
...
[NRAdude@gtmi686dpp NRAdude]$ smbmount \\slashdot.org\public -U PopeJohnPaul5 -I 64.28.67.48 -N -n TheExorcist -W slashdot.org -c mount /mnt/theposessed/public /home/NRAdude/biblestudy/virtualobjectoriented/cru cifix.c /mnt/theposessed/public
[NRAdude@gtmi686dpp NRAdude]$ mv
[NRAdude@gtmi686dpp NRAdude]$ killall samba;shutdown now;pray to god
without prejudice
Right, I am aware of mod_mmap_static, but I think that it's highly overrated at this point. Most significantly, it's at version 0.04 and placed in the "experimental" modules section for a reason: it's not well tested. /. just likes showing off. . . ;)
I think one of the reasons it's not well tested is that this is a pretty incovenient module to use. It makes you list every file to be mmapped, one-by-one, in the Apache config file. That's fine if you need it for, say, the slashbox icons, but it's going to be annoying to edit the server config files every time you want to add or remove a new banner ad.
I also just figured that they wouldn't be using 1 GB of RAM and 10k RPM SCSI drives if they had no performance concerns. Or maybe
--JRZ
Congrats, everyone -- the improvement is immediately noticeable and greatly appreciated!
Imagine the frustration (and I'm sure you can) of sitting on a T1 connection and having to stick /. in the background while it loads ... eventually.
I suppose we can safely assume it's not running IIS on an NT4 box ... :)
ikaros, who has really got to get off his bum and get his own server set up
You're only as young as the last time you changed your mind -- Timothy Leary
is there old slashcode in place? A while back, the term "declined" replaced the arguably harsher "rejected", when a story submission was turned down.
It appears it now says "rejected" once again.
Or will there be another upgrade necessary to implement those changes when ready? We're looking at replication at work right now and it's much more work than just the database itself, so I would hope this upgrade has such future developments in mind. Just curious..
The new IP is 64.28.76.48 From where I am (Melbourne, Australia) this has a ping of approx. 800ms (ATM). This is in comparison to the old IP, which took around 1000ms, a saving of 20%! Good work, guys!-
------------------------------------------
- security focus is taking about the Netpliance i-opener
- hollywood bitchslap has a link to the review of keeping the faith
- linux.com tuneup has some old tip bash history
- geek in space is on "Mellow Trancey Version"
btw, extrans (html tags to text) doesnt work (at least in preview mode)Zetetic
Seeking; proceeding by inquiry.
Elench
A specious but fallacious argument; a sophism.
It's an axiom: when a web site is good AND fast, it attracts more and more viewers. There are no good fast web sites; there are only popular, good but slow websites. I expect Slashdot to be fast for about a month, at which point it will be slow as it ever was.
At least a lot more people will be reading, though.
Oh, come one guys, Katz isn't that bad...he's a great author, even if he misses the finer technical points. The guy watches out for the geeks. Yet you spurn him.
Help us build a better map!
No wonder it's running faster...
64.28.67.48 is running Microsoft-IIS/5.0 on Windows 2000
"Do you think we could wipe out world hunger forever if scientists figured out how to make AOL's Free CD's edible?"-
P.S. I'm definitely noticing a nice speed improvement... good work guys! :-)
For the power and security a site like slashdot needs, I am surprised that they haven't switched to Microsoft Windows NT. The boys from Seattle have really done it this time: the all new version for 2000 is hard to beat! Security, stability, scalability, performance-- it's the whole package!
I know it's tempting to go for one of those fly by night 'shareware' operating systems like LinusOS. But, come on now. Slashdot is a big, grown up site. We need NT! Besides, if they get enough business, maybe the Department of Justice won't close down the internet.
Anyway, just another thought from cyberspace-- I'll sign off for now. Gotta get Outlook working again. I've been having trouble ever since I got that joke email-- hope it isn't a virus. If a virus can get past Windows security, can you imagine how many viruses infect LinusOS? I shudder to think.
So will this move make my downloading from warez.slashdot.org faster?
Oh no, looks like the two image servers are running Apache too. Sigh. I love Apache as much as the next guy, but its speed for static files is pretty bleak. /. isn't using such an add-on (since it reports no modules in its HEAD response), we're lead to assume that the server is actually going to the filesystem, loading the image into RAM, and then sending it out over the wire for every single request. Wow. Please, correct me if I'm wrong, becuase I hope slash isn't burning that many cycles. . . /. is using in their new servers that doesn't show up on the radar. Anyone know for sure?
I'd guess that these guys are serving a very limited set of files: the section images (say, 2 kb each * 60 or so images) + whatever handful of ads happen to be rotating at any given time (let's say 10 kb each * 25 active ads) + 100 kb for a few random images I forgot = 470 kb.
Even if that's a wild under-estimate, it's still clear that all the active image content at any given time could fit in a tiny fraction of the server's RAM. Apache, however, doesn't cache static files without an add-on module. So, assuming
You might want to try SGI's QSC (quick shortcut cache), which is an Apache patch (not really a module) designed for SPECweb96, or phhttpd, which is quite similar, but a bit more general-purpose. You could certainly experiment with other web servers too, but I'm assuming it's simpler to administer apache across the board.
Oh well, maybe I'm just missing some neat trick that
--JRZ
My question is, why the big deal ArrowPoint switch? That's a $15,000 unit, and it looks like all you are doing with it is firewalling and load balancing a 100Mbit pipe across 6 web servers. Sure seems to me another VA Debian box could do that for a whole lot less.
Of course ArrowPoint is in the same building as Andover.net, and Exodus is a big ArrowPoint customer. But those aren't reasons to use an expensive, closed solution when an open one would do? Are they?
No troll here, just curious.
--Seen
"I used to be a dilettante. Then I thought I'd try something else for a while."
I'm using build 2000050709; just trying at access http://www.slashdot.org/index.pl causes that mozilla build to silently crash. Interestingly enough, m15 (build 2000041811) won't load the new slashdot page at all; it just sits there, proclaiming the page is completely loaded, when the canvas is gray.
Is anyone else seeing this?
More importantly, is this just backlash for that last mozilla story from AOLTimeWarnerNetscapeICQNullSoft, or is this a Slashdot bug? ;-)
To log in, I had to set Netscape to accept cookies from anywhere (not just originating server), and I got redirected over to beta.slashdot.org. Methinks something still is a little funky.
www.eFax.com are spammers
how many people out there realize the amount of work that goes into not being able to notice any differences?
how many more do not give any appreciation to the bump in speed?
humph. Stuff looks good so far guys. Congrats on getting the job right this far.
I'm a busy guy. I've been reading this site for a few years now, and I usually set my filtering at "2" and pick through just the stories that interest me, to rein in the total amount of time I spend slashdotting. While I'm often tempted to post a comment, I've only done it a couple of times. Maybe it's fear of flame, maybe it's laziness, maybe it's just because after reading *all* the other comments to avoid replicating ideas I become so confused that I forget what triggered my original impulse to post my own comment in the first place. So my slashdot karma is usually zero.
While I appreciate that participation is rewarded in the form of moderator points, that system has a few drawbacks. My proposal is to let users select from among more than one moderation model (the one Rob gives us) on the slashdot preferences page. The preferred model for me would let every user (even lurkers like me) give a 1 (hated it), 2 (ehh, default, unrated), or 3 (loved it) rating to every comment they read, should they feel the impetus to click on the thumbs up or thumbs down button that appears below the comment. Comments could be sorted by the total accumulated points. This would permit other views, such as "top 10 hottest comments", as well as making us lurkers feel more welcome. Well?
never ask a question you don't want to know the answer to
What'd you do different? You broke the slashdot section of wmheadlines, now they're all pointing to local files...
Hey, it feels faster to me, but is that perhaps just because most people haven't caught up with the DNS changes and are going somewhere else? Have we lucky few just got a private slashdot server?
:)
I'll reserve judgement until normal load is restored.
But hey - looks like a smooth changeover. Well done.
Wasting your time since 1997.
The data format of http://slashdot.org/slashdot.xml has been changed! It broke my slashdot headlines generating script, becuase the <url> tag does not have the full URL!!
Please Fix..
Comment removed based on user account deletion
That and stats on hits/sec, SQL requests/sec, blah blah...
Congratulations on moving Slashdot.org to the new server machines.
:-(
The ultimate test will be later today, when at the middle of the day will Slashdot.org be able to keep up with the big demands from users. This was a BIG problem with Slashdot.org on weekdays, because in the middle of a weekday Slashdot.org often slowed to a deadly crawl.
Raymond in Mountain View, CA