It is probably a feature of solr, that number of documents cannot exceed 2147483519. Which, in the case of webgraph is done quite easily.
Possibly, it might by solvable by splitshard, as described here. But in that case, we need to use solr in ‘cloud mode’, is that so?
@orbiter, what is the exact difference between using ‘webgraph’ and and ‘citation reference’? I couldn’t find anywhere in the documentation…
I use an external solr (so the db process is independent on yacy and doesn’t fill the ram and other hardware constraints so easily), is that a reason? Does it happen in embeded solr as well? Did someone else meet this bug as well? What could be a solution? Could some sort of automatic split help?