Date: Mon, 3 Mar 2014 21:11:28 +0200 From: Mikolaj Golub <to.my.trociny@gmail.com> To: Andy D'Arcy Jewell <andydj@icritical.com> Cc: freebsd-fs@FreeBSD.org Subject: Re: Is it feasible to run HAST with only ONE node long-term Message-ID: <20140303191127.GA9602@gmail.com> In-Reply-To: <20140303102248.6686617b@hyperion> References: <20140303102248.6686617b@hyperion>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Mar 03, 2014 at 10:22:48AM +0000, Andy D'Arcy Jewell wrote: > My main worry is that HAST performance might suffer as the delta grows > between the master and it's long-lost slave; does the master > consume extra resources (memory, disk) "saving up" the replication > stream objects, or does it just rely on a time/version stamp so when > (in normal operation) the slave gets back in contact with the master, > it knows to do a full sync? No performance degradation is expected. hastd(8) maintains a map of dirty extents. The map size is static, just more blocks are marked as dirty during disconnect. It would only affect synchronization time after the connection is restored. You might consider setting the remote address to "none" (see hast.conf(5)) to make hastd(1) even not try to connect to the secondary. -- Mikolaj Golub
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20140303191127.GA9602>