|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] iSCSI: unsolicited Vs immediate; restart delayJulian, >To: ips@ece.cmu.edu >cc: >Subject: Re: iSCSI: draft04 questions > > > > >Julian, > >Thanks. I am not clear on some. Comments. > >>2. I didn't find a way for an iSCSI target to say that it does not support >>any unsolicited data at all. SPC-2 specifies that a zero FirstBurstSize >>means unlimited. >> >>+++ UseR2T=yes (default) and ImmediateData=no +++ > >I am confused. I thought this combination disallows just the "immediate" >data, and not the unsolicited data as a whole. The draft hints at this >in section 1.2.5. > "A target MAY separately enable immediate data without > enabling the more general (separate data PDUs) form of > unsolicited data." >=== for this case you have UseR2T=yes and ImmediateData=yes >If I misunderstood, could you please comment how immediate data alone is >disabled, while allowing unsolicited data of FirstBurstSize? >=== UseR2T=no ImmediateData=no But that puts the target in unsolicited data mode not requiring R2Ts at all! Sorry if I seem too slow. Let me try again. There are three variables - 1. solicited data mode after FirstBurstSize UseR2T=yes unsolicited data mode only UseR2T=no 2. Immediate data allowed ImmediateData=yes Immediate data disallowed ImmediateData=no 3. Unsolicited burst (immediate & separate) allowed ?? Unsolicited burst not allowed ?? FirstBurstSize can be used for (3), but a zero FirstBurstSize means "unlimited" than "not allowed". My original question was how one can distinguish the two. I would be glad to be corrected, if I am misinterpreting the usage of UseR2T. >Do I take it then that the draft currently doesn't specify the maximum time >the targets should keep the session & connection records around hoping >for a restart? I would strongly recommend adding that as an additional >field in the payload, since that is a resource allocation and scalability >issue. > >=== I assume that a traget can figure out by itself after a while that >there is no rendezvous - but I am open to requests There's no architected mechanism to figure out on its own! Either it has to keep the (session, connection, task) states around for a long time, OR it can clean up the these states and trigger an unnecessary ULP recovery on the initiator which is surprised at an unexpected restart failure (keep in mind that initiator may not be able to restart a login precisely after the "minimum time" specified, typically there's an aggregation of multiple O/S timer requests into one master handler). Providing an upper limit is a cleaner, safer design for both initiator and target. Thanks. -- Mallikarjun Mallikarjun Chadalapaka Networked Storage Architecture Network Storage Solutions Organization MS 5668 Hewlett-Packard, Roseville. cbm@rose.hp.com
Home Last updated: Tue Sep 04 01:05:26 2001 6315 messages in chronological order |