don't know if this issue came up already - in case it did and has been
dismissed, I beg your pardon. In case it didn't...
I hereby propose, that pbzip2 (https://launchpad.net/pbzip2) is used
to compress the xml dumps instead of bzip2. Why? Because its sibling
(pbunzip2) has a bug bunzip2 hasn't. :-)
Strange? Read on.
A few hours ago, I filed a bug report for pbzip2 (see
https://bugs.launchpad.net/pbzip2/+bug/922804) together with some test
results done even some few hours before that.
The results indicate that:
bzip2 and pbzip2 are vice-versa compatible each one can create
archives, the other one can read. But if it is for uncomressing, only
pbzip2 compressed archives are good for pbunzip2.
I propose compressing the archives with pbzip2 for the following
1) If your archiving machines are SMP systems this could lead to a
better usage of system ressources (i.e. faster compression).
2) Compression with pbzip2 is harmless for regular users of bunzip2,
so everything should run for these people as usual.
3) pbzip2-compressed archives can be uncompressed with pbunzip2 with a
speedup that scales nearly linearly with the number of CPUs in the
So to sum up: It's a no loose and two win situation if you migrate to
pbzip2. And that just because pbunzip2 is slightly buggy. Isn't that
Dipl.-Inf. Univ. Richard C. Jelinek
PetaMem GmbH - www.petamem.com Geschäftsführer: Richard Jelinek
Human Language Technology Experts Sitz der Gesellschaft: Fürth
69216618 Mind Units Registergericht: AG Fürth, HRB-9201
Cristian Consonni, 11/11/2015 15:09:
> I am working with a student on scientific citation on Wikipedia and,
> very simply put, we would like to use the pageview dataset to have a
> rough measure of how many times a paper was viewed thanks to
> The full dataset is, as of now, ~ 4.7TB in size.
> I have two questions:
> * if we download this dataset this would entail, from a first
> estimation, ~ 30 days of continuous download (assuming an average
> download speed of ~ 2MB/s, which was what we measured over the
> download of a month of data (~ 64GB)). Here at my University (Trento,
> Italy) this kind of downloads have to be notified to the IT
> department. I was wondering if this would be a useful information for
> the WMF, too.
No need to notify such small downloads.
> * given the estimation above I was wondering if it is possible to
> obtain this data over FedEx Bandwith (cit. ). i.e. via shipping of
> a physical disk, I know that in some fields (e.g. neuroscience) this
> is the standard way to exchange big dataset (in the order of TBs).
This assumes that some point of the network has faster download from
that machine. The server is very slow for pretty much anyone except rare
exceptions (https://phabricator.wikimedia.org/T45647 ), possibly even
inside the cluster. Copying to a hard drive might take many days.
You have two more alternatives:
* scp from Labs, /public/dumps/pagecounts-all-sites/ (sometimes reaches
3-4 MB/s for me);
* archive.org for pagecounts-raw
https://archive.org/search.php?query=wikipedia_visitor_stats (you can
start download of all months at once and use torrent, will hopefully
saturate your bandwidth because you will download from dozens servers
rather than one).
> Thanks in advance for your help.
> [*] I know these are pageviews and not unique visitors, furthermore
> there is no guarantee that viewing a citation means anything. I am
> approaching to this data the same way "impressions" versus
> "clicktroughs" are treated in the online advertising world.
>  https://what-if.xkcd.com/31/
> Wiki-research-l mailing list
If you are a user of the dewiki and frwiki dumps, please provide your
input on the proposal for splitting the dewiki and frwiki dumps into
smaller pieces, similar to that of enwiki. The proposal is available
on Phabricator.  All comments are welcome!
Also, as a reminder, please provide your comments for the next
generation of dumps on Phabricator.  This is your chance to propose
changes to the dumps to suit your needs.