From mboxrd@z Thu Jan 1 00:00:00 1970 From: Zdenek Kabelac Date: Thu, 06 May 2010 10:22:15 +0200 Subject: [PATCH 05/23] Replicator: suspend/deactivate replicator In-Reply-To: <20100506014505.GM27027@agk-dp.fab.redhat.com> References: <20100506014505.GM27027@agk-dp.fab.redhat.com> Message-ID: <4BE27C37.1040704@redhat.com> List-Id: To: lvm-devel@redhat.com MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Dne 6.5.2010 03:45, Alasdair G Kergon napsal(a): > On Wed, Apr 28, 2010 at 02:27:24PM +0200, Zdenek Kabelac wrote: >> int activation_priority; /* 0 gets activated first */ >> + int suspend_priority; /* 1 gets suspend first */ > > Can we make those two work in similar, not opposite, ways? Idea behind 'suspend_priority' is not meant as a generic concept - it's rather designed for needs of Replicator target (also that's why original reject name had the replicator in it and after your suggestion a more generic name is used here) - if there will be some future target with similar logic - we could try to figure out probably better logic. Meanwhile - the difference from activation_priority is there are no levels, it's rather boolean 'true/false' logic (which could go to comment probably) - when false==0 means default - no priority for parent suspend - thus unless explicitly set by dm user (which currently is made only by the Replicator target) the function _suspend_parent() will have no effect on processing. Zdenek