-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
hi,
in order to increase collaboration between users on the Toolserver, and
discourage single-person ownership of tools, we would like to introduce
multi-maintainer tools on the Toolserver. these are similar to stable server
projects, except there is no need to migrate to a new server; they run on the
normal Toolserver. (the flip side of this is that problems on the normal
Toolserver will affect multi-maintainer tools as well as normal tools.)
a multi-maintainer tool runs under a dedicated Unix account, which its
maintainers may access using 'sudo'. only one tool is allowed to run under
such an account, and the account may not be used for anything else.
eventually we plan to let users create these tools automatically; until then,
they must be created manually by an admin. if you would like a
multi-maintainer tool project, please file a request in the TS project in JIRA,
listing:
* your Toolserver username
* the account name for the tool
* the account names of any other users who should have access.
it may take a few days to create the account, since we're still working out
exactly how these tools will work.
if you like, you may create a multi-maintainer tool with only a single
maintainer. this will allow you to add additional maintainers later.
at the moment, there is no advantage to using a multi-maintainer tool; however,
we are looking at ways to encourage users to create such tools, and discourage
tools owned by a single user.
- river.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (HP-UX)
iEYEARECAAYFAkpshaUACgkQIXd7fCuc5vK8YQCeNksaOXyubU50ceTDdsTdF5xJ
gS4AoIOGn22IA9wXhsqSkcqPb/lTDhvh
=qfE3
-----END PGP SIGNATURE-----
Re: Rivers comment, AWB is undergoing Major changes currently for full Api
based editing.
According to a recent mono report [1], AWB is a lot closer to being Mono
with some of the dependences that we used before. We are still using the Web
Control/Browser thing for HTML rendering. The rest we would be able to work
around as appropriate..
However, saying that, AWB is still an application with a GUI.
The AWB Database Scanner code could quite easily be copied from the SVN and
reused in a small CLI application to do what you wanted.. Probably without
too many changes
Sam
[1]: http://toolserver.org/~reedy/awbmono/output.html
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
hi,
the problems caused by (we think) the failed DIMM in the NFS server seem to be
getting worse. i will be looking at an interim solution for this tomorrow, but
in the meantime the Toolserver will be fairly unreliable/slow.
we have been told the DIMM will be replaced "next week maybe"; until then there
isn't much we can do to fix the actual problem, or look for a different cause.
- river.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (HP-UX)
iEYEARECAAYFAkponGwACgkQIXd7fCuc5vJQagCeOiHc2pL2grM6ZaHpaDYlh8op
Lr8AoIVDyd7xzsmhR2CAUvkrSrRXV+pN
=FFvz
-----END PGP SIGNATURE-----
The toolserver status page (http://status.toolserver.org/) indicates
"enwiki-db defect". I could not find any more information about what the
problem is, so was wondering if anyone on the list had more information ?
I ask because one of my script has suddenly started to show strange
results (e.g.: it lists articles that have been deleted 2 weeks ago and
should not appear in results anymore); I thought I'd check here about
this "defect" before looking in detail at possible causes (such as a
schema change which would make my SQL requests incorrect, or anything
else -- but I doubt it would be the case, since my SQL requests are
copied from Mediawiki and they haven't been updated recently).
Thanks in advance !
Frédéric
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
hi,
there was an unscheduled outage of hyacinth (NFS) and stable at around 11:50
UTC. this was most likely caused by the bad memory in the server. total
outage time was under 10 minutes.
- river.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (HP-UX)
iEYEARECAAYFAkpkW7sACgkQIXd7fCuc5vKQeQCgqmzTKWDwjnpxsTQXsJhB3pft
w9sAnRXb76FYnSbrEAE4kj1TvgfCwCA/
=dwIu
-----END PGP SIGNATURE-----
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
hi,
due to an unknown issue with yarrow's disk array, the s3 database is currently
offline. the buggy firmware on these arrays makes it impossible to diagnose
the problem remotely, so s3 will remain offline until we can get someone from
Wikimedia to visit the colo. i don't know how long this will take (we have
been waiting a few weeks for some other urgent hardware maintenance...)
- river.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (HP-UX)
iEYEARECAAYFAkpi/IIACgkQIXd7fCuc5vLIgwCfehwLcoMtgJTLC6/futlJmoS9
rnIAoLlzJcJZyfyIM4mLZQF4XylHXPSw
=ULYP
-----END PGP SIGNATURE-----
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
hi,
now that stable has been moved to hyacinth, the old stable server (willow) is
idle, and has been turned into a login server. please feel free to move your
tools, if you like; however, be aware that the server might be rebooted or
reinstalled in the next few days as we fine-tune the configuration. (if it is
reinstalled, i will back up crontabs first.)
unlike the current login server (nightshade), willow is running Solaris. i
will document the implications of this on the wiki later, but the most
important differences are:
* some software is installed in different places; see
<https://wiki.toolserver.org/view/Solaris_software>
* the default $PATH places GNU tools (e.g. ls) before Solaris ones, since most
of our users are more familiar with GNU. if you would rather use the Solaris
tools, simply change your $PATH in your shell rc file.
* crontab doesn't accept the '/' syntax, so instead of writing */10, you need
to write 10,20,30,40,50.
if there are any problems (e.g. missing software or things not working right),
file a bug in JIRA, under Toolserver: willow.
at this time there are no plans to change anything on nightshade, so you can
continue to use that server if you want. (however, as willow is idle, you
might find it a lot faster, at least for now...)
- river.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (HP-UX)
iEYEARECAAYFAkpjBoIACgkQIXd7fCuc5vKqcACfRqkgaajerjhjiJnQS4XTgnNn
cX8AoIQbUvwhlboJWS2pz1/sDyiiP8Vi
=g2TZ
-----END PGP SIGNATURE-----
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
hi,
there was an unscheduled outage of the NFS server (hyacinth) and stable at
07:50 UTC today, caused by an idiot admin rebooting the wrong server by
accident. the outage time was less than 5 minutes.
- river.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (HP-UX)
iEYEARECAAYFAkpi0U8ACgkQIXd7fCuc5vIH9wCgk5R0gQVhLwi+uF8tdCCHAmtC
pZEAoIb6Lg9iTrLMSUyqcKQkopW215kC
=eACK
-----END PGP SIGNATURE-----
Cassini is back up now thanks to the Toolserver admins, since it's
inactive at the moment I'm going to set it up with a rendering stack
that'll run until Wikimania, of course if the database server pops
back up at the beginning of august we may actually have something
production ready by that time.
So what I'm going to do with it (pending objections) is:
* Set up a Planet.osm -> Cassini mirror
* Set up a multilingular PostGIS database (see
https://bugzilla.wikimedia.org/show_bug.cgi?id=19668). This should
take around 50GB of space from the 345G /sql partition
* Set up mapnik 0.6 with a multilingular CGI export script, something
like this: http://svn.openstreetmap.org/sites/tile.openstreetmap.org/cgi-bin/export
* A corresponding mod_tile / renderd config:
http://wiki.openstreetmap.org/index.php/Mod_tile
* A squid in front of this all to take the load off apache/the export script
Then I'm going to set up a MediaWiki installation on my personal box
which'll use these services via the SlippyMap plugin.
On Mon, Jul 13, 2009 at 2:12 PM, Ævar Arnfjörð
Bjarmason<avarab(a)gmail.com> wrote:
> Here's an update on the server status from Daniel, looks like we won't
> be able to do anything with the servers until the end of the month.
> Cassini (the osm-toolserver) is also down for what looks like until
> the end of the month.
>
> I'll be afk for almost all of august leading up to Wikimania so I the
> things I need before that time I'm just going to set up on my own
> somewhat more limited colo server.
>
> ---------- Forwarded message ----------
> From: Daniel Kinzler <daniel.kinzler(a)wikimedia.de>
> Date: Mon, Jul 13, 2009 at 1:19 PM
> Subject: OSM server status
> To: Ævar Arnfjörð Bjarmason <avarab(a)gmail.com>, Sven Anders
> <sven(a)anders-hamburg.de>, Kolossos <tim.alder(a)s2002.tu-chemnitz.de>
>
>
> Hi all
>
> Jens just told me that we'll have to wait some more until the OSM boxes are up.
> There is currently no console access, so Jens can't set up the OS. This will
> have to wait until Mark is back, it seems. Mark is graduating, so we only have
> someone for emergencies at the colo. Any hardware setup has to wait until he is
> back, which will be at the end of july.
>
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
hi,
at some time this morning (around 4AM GMT) there was a cooling issue at our
colo which caused zedler (sql) to shut down. unfortunately no admins were
around to fix it, so the total outage time was around 12 hours. the server is
now up.
- river.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (HP-UX)
iEYEARECAAYFAkpfTMEACgkQIXd7fCuc5vIexQCfWid297Suqhumd1bXrHEtK6X5
9jUAn23Brid2+l+U03iT6OgfdBH2VLzi
=k9FL
-----END PGP SIGNATURE-----