#89: get-entries: "end" greater than "tree_size" should be allowed

 There could be skew between frontends of a log server that uses a
 distributed implementation, which means that a get-sth request could
 return a certain tree size, but that a follow up get-entries request could
 go to another frontend that is slightly behind, and not have all the
 entries necessary.

 At the minimum, this bit of section 4.7 should be removed:

 > The "start" and "end" parameters SHOULD be within the range 0 <= x <
 "tree_size" as returned by "get-sth" in Section 4.3.

 That paragraph already somewhat conflicts with this later paragraph in the
 same section:

 > Because of skew, it is possible the log server will not have any entries
 between "start" and "end".  In this case it MUST return an empty "entries"
 array.

 This should be clarified that a log server might not have *all* the
 entries (it might still have some of them). If a subset is returned, I
 feel that the entries should still be sequential and begin with the entry
 specified by "start", of course.

 Looking at some of the new methods designed to work reduce the impact of
 possible skew (like "get-all-by-hash"), one could imagine a new method
 that returns both entries and the latest STH, so that whatever is obtained
 can be verified.

 For example, if "get-sth" gave a tree size of 10, and the client then
 issues a "get-entries" with a "start" of 5, and an "end" of 9, but the
 frontend this request goes to only has a tree size of 8, it could return
 entries 5 to 7, as well as an STH with a tree size of 8, so that those
 entries could be verified.

-- 
-------------------------+-------------------------------------------------
 Reporter:               |      Owner:  draft-ietf-trans-
  [email protected]     |  [email protected]
     Type:  defect       |     Status:  new
 Priority:  major        |  Milestone:
Component:  rfc6962-bis  |    Version:
 Severity:  -            |   Keywords:
-------------------------+-------------------------------------------------

Ticket URL: <http://trac.tools.ietf.org/wg/trans/trac/ticket/89>
trans <http://tools.ietf.org/trans/>

_______________________________________________
Trans mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/trans

Reply via email to