task mgmt clarifications
T.Krishnanand <[email protected]>
| Newsgroups | gmane.ietf.ips |
|---|---|
| Message-ID | <[email protected]> |
Hi, a) While searching thru the archives I found a discussion about the "TM response" code for abort task....i didnt see any conclusion on that ... so ..heres the question again Since the SCSI layer has only one response code "function complete" .... why should the target return "task does not exist" for an abort task ? In other words how is the "task does not exist" TM status useful to the iscsi initiator while scsi isnt interested in it .. .must be something to do with cmdSN handling ? b) A related question .... Does it matter if the target returns "function complete" for an abort task but actually it has successfully executed the cmd at the backend (ie the cmd could not be aborted)...and also makes sure that the completed task doesnt generate any response after the TM response. c) consider the following scenario: sessionA and sessionB are accessing the same LUN... 1. sessionA causes a LUN reset while sessionB is still pumping cmds. 2. sessionB issues an "abort task" for a task ... which happens to be the first task in sessionB to be executed after the LUN reset (and hence carries the rest UA) suppose the cmd has executed in the backend (before it could be aborted) and has thus cleared the reset UA on sessionB ... should the abort task be failed with say "task does not exist" so that the scsi initiator comes to know of the LUN reset ? or is it ok to abort the cmd carrying reset UA .. in this case the initiator behind sessionB would never know about the LUN reset. cheers, tkrishna _______________________________________________ Ips mailing list [email protected] https://www1.ietf.org/mailman/listinfo/ips