From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Date: Thu, 26 Aug 2004 10:27:50 +0200 From: Lars Marowsky-Bree To: Philipp Reisner , drbd-dev@lists.linbit.com Subject: Re: [Drbd-dev] Summary (3rd try) Message-ID: <20040826082750.GA3125@marowsky-bree.de> References: <200408251713.45325.philipp.reisner@linbit.com> Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <200408251713.45325.philipp.reisner@linbit.com> Cc: List-Id: Coordination of development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On 2004-08-25T17:13:45, Philipp Reisner said: > Hi, > > I tried to incooperate the comments of LMB and LGE. Please review: > > after-pri-pri-now-sec-sec = I wish I had a better name here ;-) > disconnect No automatic resynchronisation gets performed. One > node should drop its net-conf (preferable the > node that would become sync-target) > DEFAULT. Hmmm. Should they drop the netconf, or should instead the first node to be elected primary become the SyncSource implicitly? > ass-older Auto sync source is the oder primary (curr.behaviour i.t.s.) > ass-younger Auto sync source is the younger primary I'm a bit concerned about the older vs younger distinction. "before" and "after" may be slightly better at conveying what you mean. > ass-furthest Auto sync source is the node that did more modifications > ass-NODENAME Auto sync source is the named node > pri-sees-sec-with-higher-gc = > disconnect (current behaviour) > ass-primary Auto sync source is the current primary > panic The current primary panics. The node with the > higher gc should take over. Seems ok. > Notes: > 1) The disconnect actions cause the sync-target or the secondary > node to go into StandAllone state. > 2) If two nodes in primary state try to connect one of them goes > into StandAllone state (=curr. behaviour) This is not quite symmetric, which means special cases ;-) Shouldn't they either refuse to connect, or both drop to StandAlone? > 3) As soon as the decission is takes the sync-target addopts the > GC of the sync source. > [ The whole algorith would also work if both would reset their > GCs to <0,0,0...> after the decission, but since we also > use the GC to tag the bitmap it is better the current way ] Ok. > Is this better than try2 ? > Should we improve the naming ? "ass" might not be the best choice. auto-sync-from- > Do you think that the naming is constent ? > Is it still ambiguous ? I hope it's not, but the other Lars is better at spotting inconsistencies than I am ;-) Sincerely, Lars Marowsky-Brée -- High Availability & Clustering \\\ /// SUSE Labs, Research and Development \honk/ SUSE LINUX AG - A Novell company \\//