On 03/12/2012 09:33 PM, no-re...@cfengine.com wrote: > - Is there a reasonable way to port existing information from the BDB > databases to the new Tokyo Cabinet databases? Presuming no change in the > actual data being stored, the "learn what's normal on the server" data would > be handy to start off with rather than having every system start over from > scratch. This is certainly not an insurmountable issue either way, but, you > know, if it's easy and possible to retain current data (particularly > cf-monitord stuff)... :)
Good question. Turned out, it is not that hard. I have just committed simple shell scripts which convert from Berkeley DB to Tokyo Cabinet and QDBM databases by using corresponding command-line tools: https://github.com/cfengine/core/blob/master/contrib/berkeley2tc https://github.com/cfengine/core/blob/master/contrib/berkeley2qdbm I have only tried it on a few databases I had around, so bugfixes are welcome. > - Is it safe to presume that the precompiled binaries will continue to come > with the library pre-built Yes, packages distributed from cfengine.com will continue to bundle all the dependencies. > or should I get on top of packaging libTokyoCabinet for my Linux systems now? Only if you build CFEngine yourself. > I suppose the "why not SQLite?" question should be tossed in here just for > fun, too. Actually, experimental SQLite backend was removed together with Berkeley DB. https://github.com/cfengine/core/commit/e6f094c2a7130d7db2259e0891b6cf13be3c3232 > Is it slower than other options? Yes, dramatically. -- Mikhail Gusarov _______________________________________________ Help-cfengine mailing list Help-cfengine@cfengine.org https://cfengine.org/mailman/listinfo/help-cfengine
