We're still afloat so far. :)
It performed admirably; I was amazed the server stayed up at all. Most can't handle a slashdotting; I'm sure this is testament to LDC's coding skill and your foresight in turning off some of the more demanding processes.
I wonder about the server now, though: wikipedia seems to be down; it won't bring up Recent Changes in a new window and the edit window I have open is hung on perpetually "loading." Are we being slashdotted again (more heavily!) or is there some other cause?
cheers,
kq (circa 6:00 UTC)
On ĵaŭ, 2003-01-23 at 22:06, koyaanisqatsi@nupedia.com wrote:
We're still afloat so far. :)
It performed admirably; I was amazed the server stayed up at all. Most can't handle a slashdotting; I'm sure this is testament to LDC's coding skill and your foresight in turning off some of the more demanding processes.
I wonder about the server now, though: wikipedia seems to be down; it won't bring up Recent Changes in a new window and the edit window I have open is hung on perpetually "loading." Are we being slashdotted again (more heavily!) or is there some other cause?
I tried lifting the ban on updating view counters and the slow special pages and load ballooned up like a, well, a balloon. I put it back.
So we're still not out of it yet. :)
-- brion vibber (brion @ pobox.com)
On 23 Jan 2003, Brion Vibber wrote:
On ĵaŭ, 2003-01-23 at 22:06, koyaanisqatsi@nupedia.com wrote:
I wonder about the server now, though: wikipedia seems to be down; it won't bring up Recent Changes in a new window and the edit window I have open is hung on perpetually "loading." Are we being slashdotted again (more heavily!) or is there some other cause?
I tried lifting the ban on updating view counters and the slow special pages and load ballooned up like a, well, a balloon. I put it back.
Part of the problem might be MySQL. It is very fast for applications that are entirely static, however, when the data are being changed, it is quite slow. In particular, MySQL scales very poorly to multiple users trying to simultaneously update the database. I see from the statistics that the typical Wikipedia page is read only 4 or 5 times for each time it is updated. For such a use profile, PostgreSQL is likely to offer better performance. Of course, the only benchmark that counts is the application at hand. Has anyone tried Wikipedia on PostgreSQL?
Apologies if I've strayed off topic, M Carling
On ĵaŭ, 2003-01-23 at 23:18, M Carling wrote:
On 23 Jan 2003, Brion Vibber wrote:
I tried lifting the ban on updating view counters and the slow special pages and load ballooned up like a, well, a balloon. I put it back.
Part of the problem might be MySQL.
We've heard this song before, and it's never yet been followed by "and here's my patch which makes Wikipedia run on PostgreSQL, which I am an expert with and that's how I know it's so damn fast, and here are some precise and useful tips on optimising the setup and converting the database."
Apologies if I sound a little rude, but it does get a little frustrating after a few times. :)
If you would like to try adapting the software to PostgreSQL, I'd *love* to see the results. Please sign up for the wikitech-l developers' list, and check our source out of CVS:
http://meta.wikipedia.org/wiki/How_to_become_a_Wikipedia_hacker
-- brion vibber (brion @ pobox.com)
On Fri, Jan 24, 2003 at 12:36:58AM -0800, Brion Vibber wrote:
On ??a??, 2003-01-23 at 23:18, M Carling wrote:
On 23 Jan 2003, Brion Vibber wrote:
I tried lifting the ban on updating view counters and the slow special pages and load ballooned up like a, well, a balloon. I put it back.
Part of the problem might be MySQL.
We've heard this song before, and it's never yet been followed by "and here's my patch which makes Wikipedia run on PostgreSQL, which I am an expert with and that's how I know it's so damn fast, and here are some precise and useful tips on optimising the setup and converting the database."
Don't get pissy, Brion. mod_wiki is en route. You won't see most of the advantages of Postgres without redesigning the database schema to take advantage of it.
Apologies if I sound a little rude, but it does get a little frustrating after a few times. :)
Yes, but not as frustrating as trying to actually dive in and make such a patch; the MySQLisms and the fact the SQL statements are created on the fly instead of isolated into one easily modifiable file, the way the Scoop codebase does, makes it very difficult.
If you would like to try adapting the software to PostgreSQL, I'd *love* to see the results. Please sign up for the wikitech-l developers' list, and check our source out of CVS:
I agree; if you have the stamina for it, please provide a patch so we can see how Postgres supports Wikipedia as it stands right now. mod_wiki is a fourth generation thing, while the current codebase is third generation.
Jonathan
On 24 Jan 2003, Brion Vibber wrote:
On ĵaŭ, 2003-01-23 at 23:18, M Carling wrote:
Part of the problem might be MySQL.
Apologies if I sound a little rude, but it does get a little frustrating after a few times. :)
You didn't come across as rude. I was just asking if anyone had tried. Seems not.
http://meta.wikipedia.org/wiki/How_to_become_a_Wikipedia_hacker
Thanks for the link. I checked it out.
M Carling