> Anyone know why the digests are failing so frequently? Are they packaged > poorly? Is it the mirror I'm getting them from? But then that begs the > question, aren't the mirrors synchronized and MD5 verified with each > other? > Shouldn't TCP/IP be safe in that it retries a packet until it gets it -- > and > therefore, why is the MD5 different?
I haven't had any issues with md5 sums on recent packages and my systems are updated pretty much daily (thus requiring frequent downloads). When it has happened I had to delete the file from /usr/portage/distfiles in order to fetch a clean copy. And since I'm running http-replicator I will also remove any cached copy and recycle http-replicator to ensure it is starting from a clean state. If you're using this or another cache/proxy you'll want to ensure that you clean it out before retrying the fetch. I'm not sure how the mirrors are synchronized but I would guess it's rsync based. There should not be any problems there. I don't know if they check the md5 sums on files pulled into the mirror, but I would guess they do not. I'd look at your /var/log/emerge.log for errors dealing with the fetch. Also check for filesystem errors if you're using a separate FS for the /usr/portage space; it could be a read/write error associated with the disk itself. -- gentoo-user@gentoo.org mailing list
