From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:39679) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1eqnFU-0006ps-5v for qemu-devel@nongnu.org; Tue, 27 Feb 2018 16:57:25 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1eqnFS-0004hX-Ti for qemu-devel@nongnu.org; Tue, 27 Feb 2018 16:57:24 -0500 References: <20180223235142.21501-1-jsnow@redhat.com> <20180223235142.21501-20-jsnow@redhat.com> <2c7f82a8-8b19-4f2d-b6d5-3c77258edf91@redhat.com> From: John Snow Message-ID: Date: Tue, 27 Feb 2018 16:57:01 -0500 MIME-Version: 1.0 In-Reply-To: <2c7f82a8-8b19-4f2d-b6d5-3c77258edf91@redhat.com> Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [RFC v4 19/21] blockjobs: Expose manual property List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Eric Blake , qemu-block@nongnu.org Cc: kwolf@redhat.com, pkrempa@redhat.com, jtc@redhat.com, qemu-devel@nongnu.org On 02/27/2018 03:16 PM, Eric Blake wrote: > On 02/23/2018 05:51 PM, John Snow wrote: >> Expose the "manual" property via QAPI for the backup-related jobs. >> As of this commit, this allows the management API to request the >> "concluded" and "dismiss" semantics for backup jobs. >> >> Signed-off-by: John Snow >> --- >> =C2=A0 blockdev.c=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0= =C2=A0 | 19 ++++++++++++++++--- >> =C2=A0 qapi/block-core.json | 32 ++++++++++++++++++++++++++------ >> =C2=A0 2 files changed, 42 insertions(+), 9 deletions(-) >> >=20 >> +++ b/qapi/block-core.json >> @@ -1177,6 +1177,16 @@ >> =C2=A0 # @job-id: identifier for the newly-created block job. If >> =C2=A0 #=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 omitted= , the device name will be used. (Since 2.7) >> =C2=A0 # >> +# @manual: True to use an expanded, more explicit job control flow. >> +#=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 Jobs may tran= sition from a running state to a pending state, >> +#=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 where they mu= st be instructed to complete manually via >> +#=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 block-job-fin= alize. >> +#=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 Jobs belongin= g to a transaction must either all or all not >> use this >> +#=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 setting. Once= a transaction reaches a pending state, >> issuing the >> +#=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 finalize comm= and to any one job in the transaction is >> sufficient >> +#=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 to finalize t= he entire transaction. >=20 > The previous commit message talked about mixed-manual transactions, but > this seems to imply it is not possible.=C2=A0 I'm fine if we don't supp= ort > mixed-manual transactions, but wonder if it means any changes to the > series. >=20 > Otherwise looks reasonable from the UI point of view. >=20 More seriously, this documentation I wrote doesn't address the totality of the expanded flow. I omitted dismiss here by accident as well. This is at best a partial definition of the 'manual' property. I'd like to use _this_ patch to ask the question: "What should the proper noun for the QEMU 2.12+ Expanded Block Job Management Flow Mechanism be?" I'm not too sure, but "Manual mode" leaves a lot to be desired. I keep calling it something like "2.12+ Job Management" but that's not really descriptive. I conceptualize the feature as the addition of a purposefully more "needy" and less automatic completion mechanism, hence the "manual" Anyway, I'd like to figure out a good "documentation name" for it so I can point all instances of the creation property (for drive-backup, drive-mirror, and everyone else) to a central location that explains the STM and what exactly the differences between manual=3Don/off are. I'd the= n like to expose this property via query and link the documentation there to this description, too. It'd be nice-- under the same arguments that prompted 'dismiss'-- to say that if a client crashes it can reconnect and discover what kind of attention certain jobs will need by asking for the manual property back. --js