Below is a report for the last week on how often the [[Chemstry]] page
took more than five seconds to load, according to my script.
Response times above five seconds can be considered as down time,
since the website feels so slow that users start to abandon it.
However, no actual down time was recorded this week. (Let's recall
how phase II worked, and bless the current blitz performance.)
The [[Chemistry]] page is rather short and simple, and should be
served as fast as the network can carry it. If the server spends
several seconds on each serving of this page, it will seem very busy.
The sooner it can get the page out of its hands, the better.
Mon, Nov 11, 18%
Tue, Nov 12, 8%
Wed, Nov 13, 10%
Thu, Nov 14, 2%
Fri, Nov 15, 4%
Sat, Nov 16, 2%
Sun, Nov 17, 4%
Mon, Nov 18, 6%
Tue, Nov 19, 7%
Around the clock (UTC time) over all days:
00:00 - 01:59 8% (7-9 pm EST, 4-6 pm PST)
02 6% (9-11 pm EST, 6-8 pm PST)
04 3%
06 3%
08 0%
10 0%
12 8% (7-9 am EST, 4-6 am PST)
14 6% (9-11 am EST, 6-8 am PST)
16 15% (11 am - 1 pm EST, 8-10 am PST)
18 11% (1-3 pm EST, 10-12 am PST)
20 9% (3-5 pm EST, 12-2 pm PST)
22:00 - 23:59 18% (5-7 pm EST, 2-4 pm PST)
Conclusion: Europeans should use Wikipedia in the morning.
At noon, the Americans wake up and the site gets bogged down.
--
Lars Aronsson (lars(a)aronsson.se)
Aronsson Datateknik
Teknikringen 1e, SE-583 30 Linuxköping, Sweden
tel +46-70-7891609
http://aronsson.se/http://elektrosmog.nu/http://susning.nu/
There seems to be some interest in creating a static HTML distribution
(dump) of Wikipedia, most notably it is requested on the
Wikipedia:Database_download page and in Feature Requests #596830 on
Sourceforge. This would allow people to download the Wikipedia for use
offline, for example from a CDROM.
So, I have started work and made my initial version (English only)
available online for anyone on this list to evaluate and test. I am
looking for feedback, suggestions, bug reports and general comments.
http://www.rawlinson.ca:8080/wikipedia/index.html
Please do not attempt to mirror the site as my server and bandwidth won't
be able to handle it. The site is only intended for developers to try and
give feedback. Once everyone is happy with it I will make .tar and .iso
packages available for distribution.
At the moment, the method I use to create the static HTML version is very
lengthy in terms of processing time and requires a number of manual step.
It takes my 1 GHz machine about 5 hours to generate all the pages.
Ideally, I'll have something more automated and efficient as time goes on.
My plan for how things would work is I will produce an updated static HTML
version every few months or significant milestones.
I'm not sure how to distribute this static HTML version when it's ready
for a public release. Currently it's about 500 Meg in size (that includes
everything). As I mentioned above I have limited server resources. For
distribution maybe it could be put on the Sourceforge download page, or on
the Wikipedia.org server somewhere (/tarballs)?
Finally, since I am new to Wikipedia and this list, please excuse me while
I learn how things work around here. I am open to criticism, suggestions
and discussion. I am looking forward to working with everyone on
Wikipedia and contributing where I can.
Some Technical Details (for those interested):
- English only (currently)
- uses "printable" pages, no top or side navigation bars
- added links to home, back, copyright and Wikipedia.org to bottom of all
pages (TODO: if a talk page exists a link should be added)
- pages are stored in directories based on first two characters of MD5
hash, same as image storage scheme
- includes all namespaces (talk, users, users_talk, wikipedia_talk, etc.)
- created a list with links to all the items in each namespace to allow
for basic searching of page titles
- redirects replaced with direct link to article
Regards,
Steve Rawlinson
I've signed up for the technical list againand now I
find myself drowning in Wikipedia mailing lists... but
next week I should have a nice computer in Slovenia to
work with, so hopefully that will not be a problem.
:)
Anyway, we now have a problem (bug) on the Esperanto
Wikipedia and I thought perhaps other people on the
list might have some ideas. Basically, Esperanto has
special non-ASCII characters that are often typed on
the Internet by using an x after the ASCII letter, but
they are usually displayed in Unicode.
We need to be able to allow our users to type in
x-system or Unicode and have the result be in Unicode.
We also need to allow the users to view the Esperanto
encyclopedia in x-system in case they are using an
older browser that does not support it. In the older
software, this worked perfectly. We also had an extra
link on the top which allowed the user to switch their
view simply from X-System to Unicode and back by
manually editing a user preference.
Is this possible simply by editing EoLanguage.php or
do will we need to edit other files as well? Which
one(s)? Also, it would help if Brion could give a
technical explanation of the problem here so perhaps
even I could try to help solve it soon. I would like
this to be fixed before I start the next wave of
promotion.
Thanks,
Chuck Smith
=====
I'm in the Czech Republic!! Mi estas en Cxehxio!!
=========================================
Travel Plans: http://eo.wikipedia.com/wiki/Chuck_SMITH
My Webpage: http://amuzulo.babil.komputilo.com/
Enciklopedio: http://eo.wikipedia.com/
__________________________________________________________________
Gesendet von Yahoo! Mail - http://mail.yahoo.de
Möchten Sie mit einem Gruß antworten? http://grusskarten.yahoo.de
Brion,
Is it possible, i.e., permissible to erase the "Edd Poor" username? It's too close to my own real name for comfort -- and it was only a vandal anyway. If you can do this, how about converting it to an IP address?
Ed Poor
Is a separate integer namespace property actually necessary? I got the
impresion that namespaces were designated by a prefix in the article
title, such as Special:, User:, or Wiki:
In that case, one could find all the articles in the Wiki: namespace by
doing this query:
SELECT article_id FROM cur WHERE title LIKE "Wiki:%" ;
And if we had some indexes defined, that search would actually be
extremely quick to do.
Any reasons why we need to keep a separate column for "namespace"?
If we want to keep namespaces in their own column, maybe we should have
a table of namespace id's and their corresponding prefix, and then NOT
put that prefix as part of the articles actual title.
Jonathan
--
Geek House Productions, Ltd.
Providing Unix & Internet Contracting and Consulting,
QA Testing, Technical Documentation, Systems Design & Implementation,
General Programming, E-commerce, Web & Mail Services since 1998
Phone: 604-435-1205
Email: djw(a)reactor-core.org
Webpage: http://reactor-core.org
Address: 2459 E 41st Ave, Vancouver, BC V5R2W2
I just learned a bit about InnoDB. While it seems to provide significant
performance advantages and reduce locking issues, it doesn't, in the
current stable version, support MySQL/MyISAM's FULLTEXT index. That's
the reason Kuro5hin currently only allows 30 day searches. According to
InnoDB.com, MySQL 4/InnoDB will support FULLTEXT.
So InnoDB isn't an option for the time being. If we want to get quick
improvements, moving the article counters out of the CUR table sounds
like a good idea.
PostgreSQL appears to have a module for full text indexing:
http://techdocs.postgresql.org/techdocs/fulltextindexing.php
Regards,
Erik
--
FOKUS - Fraunhofer Insitute for Open Communication Systems
Project BerliOS - http://www.berlios.de
Brion,
Thank you for posting the queries which each page view request triggers. These might be optimized somewhat, as you suggested.
I'm ever more concerned, though, with the pauses of 10 seconds or more -- some even lasting several minutes! -- during which no access to the Wikipedia is possible at all.
Are these "no access" periods due to the page view queries? I thought the delays were caused rather by table locks.
In my experience with MS SQL Server 6.5 (and above), mere read-only queries do not cause table locking. Rather, it is only update queries which lock tables -- especially when performed within a "transaction".
Ed Poor
Could people please check the multi-language search at
http://www.wikipedia.org/tools/wikisearch.php? I get at the moment all kinds
of search results covering one another, making the result completely
unreadable.
Andre Engels
Dear developers and sysops,
The Polish Wikipedia community would like to ask for setting up
Polish language mailing list.
We urgently need the list to help us improve our editorial process
and decision making.
We hope the new list can be open for our contributors ASAP.
Thanks in advance.
Best regards,
Kpjas.