Date: Fri, 14 Sep 2001 12:30:04 -0700 From: Marcel Moolenaar <marcel@xcllnt.net> To: Garrett Wollman <wollman@khavrinen.lcs.mit.edu> Cc: hubs@FreeBSD.ORG Subject: Re: Randomizing access to CVSup mirrors shows inconsistency Message-ID: <20010914123003.C1150@dhcp01.pn.xcllnt.net> In-Reply-To: <200109141911.f8EJB8k98075@khavrinen.lcs.mit.edu> References: <20010914120655.A1150@dhcp01.pn.xcllnt.net> <200109141911.f8EJB8k98075@khavrinen.lcs.mit.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, Sep 14, 2001 at 03:11:08PM -0400, Garrett Wollman wrote: > <<On Fri, 14 Sep 2001 12:06:56 -0700, Marcel Moolenaar <marcel@xcllnt.net> said: > > > 1. Is it actually valid to randomize access to mirrors? > > No. There is (by intention) no synchronization between the mirrors' > own updates from the master, so it is possible that two consecutive > updates from different servers will actually retrograde your > repository. I was afraid of that. I assume there's no option in the CVSup protocol that handles randomization by preventing the client from retrograding. I don't mind picking up a server that hasn't yet sync'd up to where I already happen to be, provided it can be made a no-op. I know I will sync up the next hour, so all I've lost is an hour. In more than 90% of the cases this isn't even a loss. An alternative is to set my interval to more than an hour (say 2 hours) so that I've guaranteed that all mirrors had a change to sync-up and are at least equal to me or ahead of me. This depends on the greatest interval used by the servers in the domain I picked. Can I assume that all mirrors sync once every hour? -- Marcel Moolenaar USPA: A-39004 marcel@xcllnt.net To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hubs" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010914123003.C1150>