I suspect that for development purposes we'll want to be able to independently control read-path and write-path checksumming. So I vote for (2). Allen Samuels SanDisk |a Western Digital brand 951 SanDisk Drive, Milpitas, CA 95035 T: +1 408 801 7030| M: +1 408 780 6416 allen.samuels@xxxxxxxxxxx > -----Original Message----- > From: ceph-devel-owner@xxxxxxxxxxxxxxx [mailto:ceph-devel- > owner@xxxxxxxxxxxxxxx] On Behalf Of Igor Fedotov > Sent: Friday, July 15, 2016 4:38 AM > To: ceph-devel <ceph-devel@xxxxxxxxxxxxxxx> > Subject: bluestore checksum enable/disable behavior > > Hi All, > > During read regression testing I realized that one can't disable > checksum verification on read if corresponding blob was created with > csum verification enabled. > > More detailed overview is below. > > We have "bluestore csum" flag in ceph config that currently > enables/disables csum generation for blob writing only. It appeared that > this flag has no impact on the read request handling. During blob > reading we simply check whether csum tag is attached for the blob and > verify csum if it's there. > > The question is if such behavior is OK or we need finer way to control > csum verification on read? > > I can see a couple of ways to improve that. > > 1) Use "bluestore csum" flag during the reading too. > > 2) Introduce another flag ("bluestore csum on read"?) to have explicit > control on read handling. Thus we'll have independent control on both > read and write. > > The second way is more flexible but IMHO is an overkill hence I vote for 1). > > Any thoughts? > > > Thanks, > > Igor > > > > > > -- > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html ��.n��������+%������w��{.n����z��u���ܨ}���Ơz�j:+v�����w����ޙ��&�)ߡ�a����z�ޗ���ݢj��w�f
