There are two use cases. 1) host shared data so that we can all point to and work from the same data, ideally both literal docs and also extracts (text/metadata .json files representing extracted information).
2) a modest vm to allow all of us to run the regression tests We could use help with either or both. What we had before: 8 GB RAM 8 cores 4 TB -- 2TB for docs, 1TB for extracts, 1TB for staging We can always use more RAM and more cores up to the point of I/O bottlenecks. On Tue, Jun 2, 2020 at 6:37 AM Maruan Sahyoun <[email protected]> wrote: > is that a storage box only or does it need to do some computings too? > > Maybe you could write a small spec for the server requirement? > > BR > Maruan > > > > Still haven’t had time to put the server in a dmz. Ugh. > > > > Yes, more than happy to share. > > > > If anyone has recommendations for file hosting for a couple of TB, let me > > know. > > > > One option would be to work with CommonCrawl to bump the max file size > one > > crawl a year... > > > > On Tue, Jun 2, 2020 at 1:48 AM Tilman Hausherr <[email protected]> > > wrote: > > > > > Can we / I access these files? Most differences are improvements or not > > > meaningful, but there are a few I'd like to have a look, e.g. > > > > > > commoncrawl3/commoncrawl3/XO/XOAAGISRMRPZQRZF4LSMJERGEYK5QI2T > > > > > > the word "antrag" loses the first "a". Although maybe the "a" was a big > > > one and gets assigned to another line. > > > > > > Tilman > > > > > > Am 02.06.2020 um 02:58 schrieb Tim Allison: > > > > > > Reports are available here: > > > > https://github.com/tballison/share/blob/master/tika_comparisons/reports-pdfbox-2.0.20.tgz > > > > Looks like there are trivial differences in content with a slight > > > > improvement over 2.0.19. I don't see any differences in exceptions > or > > > > attachments. > > > > > > > > Cheers, > > > > > > > > Tim > > > > > > > > > > --------------------------------------------------------------------- > > > To unsubscribe, e-mail: [email protected] > > > For additional commands, e-mail: [email protected] > > > > > > > > >
