|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] RE: iscsi : default iscsi mode page settings.It has exactly the FCP-2 meaning. Julo "Elliott, Robert" <Robert.Elliott@c To: "IPS Reflector" <ips@ece.cmu.edu> ompaq.com> cc: Sent by: Subject: RE: iscsi : default iscsi mode page owner-ips@ece.cmu settings. .edu 24-09-01 02:09 Please respond to "Elliott, Robert" > -----Original Message----- > From: Santosh Rao [mailto:santoshr@cup.hp.com] > Sent: Friday, September 21, 2001 2:27 PM > > If an initiator cares about any settings it better run MODE SENSE > > to set them. In a multi-initiator environment with various kinds (I meant both MODE SENSE and MODE SELECT) > > of reset events occuring on the SAN, an initiator can't be sure > > of the state of a mode page. (see the tape problem discovered by > > Veritas in FCP-2 public review; tape mode pages were reset without > > the knowledge of the application, breaking backups). > > Isn't this all the more reason to avoid setting FirstBurstSize > through a mode page and use a login key for this ? Shared mode > pages can cause some strange side effects on I/Os that are > using un-solicited data. If there were not already a FirstBurstSize for other protocols controlled by the mode page, I'd agree - these protocol-specific fields shouldn't be controlled by mode pages, which can be changed by applications at any time. They ought to be properties of the login session and controlled only by the iSCSI drivers. However, it has been defined this way for other protocols, so I think a compatible control is needed for iSCSI. This makes protocol bridges simpler, and will avoid confusing software that does try to use these mode pages. A page that appears when a FC target is connected to a host via FC should also appear with the same meaning when presented through an iSCSI bridge via Ethernet. As Sanjeev noted in another email, this mode page is LU specific which would be hard to emulate in the login key approach. > > > Logical Unit Control Mode Page > > > ============================== > > > ... > > ... > Does the EPDC bit verify the CRN capabilities of both the target > FCP layer and the target SCSI ULP (device server) ? One would think > the capabilities of the 2 layers need to be checked through seperate > bits. The device server's ability to support CRN is a SCSI ULP > feature and seems more like a bit in the Control Mode Page. > (does not exist today ?). The target FCP's CRN capability is > tested through the FCP specific EPDC bit in the FCP > specific lun control mode page. The CRN is carried as part of the IU (aka PDU), so I think of it as part of the iSCSI layer, not the SCSI layer. Stuff in the CDBs is what the SCSI layer deals with. Enabling or disabling CRN is done through a mode page, which is SCSI layer activity. For protocol-specific mode page fields (like EPDC and FirstBurstSize) the SCSI layer has to talk to the iSCSI layer to read/write the controls. The disconnect-reconnect mode page has fields that can apply to most different protocols like FirstBurstSize. Provided the semantics are the same, any protocol offering those features should use the mode page to control them. New issue ========= iSCSI's FirstBurstSize does appear to have one difference from that described in SPC-2 and FCP-2: iSCSI-07-90 requires the entire transfer fit in the FirstBurstSize if enabled, while SPC-2/FCP-2 have no such restriction. I'd like to see it work the same as the other protocols. iSCSI: "It is also an error for an initiator to send more data, whether immediate or as separate PDUs, than the SCSI limit for first burst." SPC-2: "The FIRST BURST SIZE field indicates the maximum amount of data that may be transferred to the target for a command along with the command." FCP-2: "...If the maximum amount of data has been transmitted, but more data remains to be transferred, the target shall request that data with subsequent FCP_XFER_RDY IUs." --- Rob Elliott, Compaq Server Storage Robert.Elliott@compaq.com
Home Last updated: Mon Sep 24 16:17:17 2001 6698 messages in chronological order |