I have extended the FAQ entry #dumpload [1], about the dump+load procedure, with some caveats and details on the technique to dump+load a large repository with minimal downtime. This is based on a users@ mail from last year [2] (I had been meaning to convert this into a FAQ since then, but only now got around to it). See my edits in r1803274 and r1803323. Questions on how to correctly execute a dump+load (possibly from a very old repository) still come up regularly on users@.
I have also written a new (related) entry #fix-nonLF-log [3], describing how to fix the "svnadmin: E125005: Cannot accept non-LF line endings in 'svn:log' property" error one might run into during 'svnadmin load'. See r1803336. These FAQs more or less cover one of the whiteboard items from the 2015 hackathon [4] (namely "document ideal procedure for full dump/load with minimal down-time"). Review, additions, rewriting, ... welcome :-). Also: this reminds me of those other whiteboard items from 2015: "svnadmin load --normalize-revprops", "- normalize svn:* props", "- normalize mergeinfo", to get 'svnadmin load' on par with svnsync regarding normalization. Having those options (at least the first two) would make those faq answers a lot shorter ... If anyone has some cycles to spare, 1.10 might be a good target to try and get these in ... [1] http://subversion.apache.org/faq.html#dumpload [2] https://svn.haxx.se/users/archive-2016-06/0098.shtml [3] http://subversion.apache.org/faq.html#fix-nonLF-log [4] https://svn.haxx.se/dev/archive-2015-09/0187.shtml -- Johan
