Gérald Macinenti <[EMAIL PROTECTED]> wrote: > ok, two advices to do it is enough, I won't resist ;) but I would like > to be able to return back to a previous state if anything goes wrong > (what my complicated solution allows), then is it possible in your > solution to vos copy instead of vos move ?
Yep, vos copy should work the same as a move. Although if someone changes the original after you've copied it, those changes obviously will not be copied to the new server. Oh, wait, I forgot. You have to pick a different name for the volume when you do a copy. You could just cal every volume "name.copy" and vos rename it back to "name" once it was on the new cell. Or you could vos addsite and then vos release. And once you only have the new server, seperate it from the old and vos convertROtoRW. > ps: when you say "migrate the databases", do you mean a copy of files > from the "db" directory? if so, what files exactly, prdb*, vldb* and > what about bdb* ? Yeah, I mean those files. Generally one uses AFS itself to replicate these files. They are automatically replicated between all running (and properly configured) instances in a cell. I'm not sure if its safe to just copy them directly or not. The data might be endian spcific. But someone who userstands source code can probably provide a definate answer on that. <<CDC _______________________________________________ OpenAFS-info mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-info
