My mistake, the algorithm is actually supposed to be to cache in the datastore only if the node is closer to the key than its peers. This fix has gone in in 938.
On Sun, Aug 06, 2006 at 01:21:11PM +0300, Jusa Saari wrote: > On Sat, 05 Aug 2006 20:14:54 +0100, Matthew Toseland wrote: > > > Build 933 is now available from SVN and from emu (update.sh/cmd). Please > > test it. It will be released to the auto-updater on monday if no serious > > bugs are reported. This build fixes some datastore bugs, and includes > > recent work, converting the store to a short term cache and a long term > > store (the existing store becomes the cache). The long-term store only > > stores inserts, and only those for which the key is closer to this node > > than to any node on the insert chain so far. > > So, if the originating host happens to be the closest for the given key in > the entire network (which, in a network of a few hundred or at most few > thousand nodes and small file chunks is going to happen quite often), it > won't be stored at all, despite the insert seemingly completing > succesfully? > > And of course, since nodes sometimes leave from the network or change > their position, even popular content's availability will deteriorate > steadily as it's permanently stored on less and less datastores. > > > 80% of the storeSize should eventually be used by the store rather than > > the cache, but to avoid major loss of content I have arranged for the > > cache to be gradually shrunk as the store grows. For a full store, this > > means that initially it will be cut by 20% and renamed to cache. If the > > node is then not restarted for a long time, it may use up to 60% over the > > storeSize limit. However, every time the node is restarted, the cache will > > be shrunk to make room for any growth in the store (up to the limits > > specified above; 20% of storeSize is always reserved for the cache), and > > if the cache is over the size limit it will not grow further. Given the > > frequency of updates, and given that the store grows slowly, hopefully > > this will not be a practical problem. This code will be removed before > > 0.7.0, but not for some time. > > > _______________________________________________ > Tech mailing list > Tech at freenetproject.org > http://emu.freenetproject.org/cgi-bin/mailman/listinfo/tech > -- Matthew J Toseland - toad at amphibian.dyndns.org Freenet Project Official Codemonkey - http://freenetproject.org/ ICTHUS - Nothing is impossible. Our Boss says so. -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 189 bytes Desc: Digital signature URL: <https://emu.freenetproject.org/pipermail/tech/attachments/20060809/95607244/attachment.pgp>
