diff options
author | Rob Austein <sra@hactrn.net> | 2007-04-04 20:01:24 +0000 |
---|---|---|
committer | Rob Austein <sra@hactrn.net> | 2007-04-04 20:01:24 +0000 |
commit | ece6c7171ca5dca62a4ae71666b8e15b2dea8799 (patch) | |
tree | 621f76f996005b9b9831e77e488a1c1969ec68d2 /docs/repository-structure.txt | |
parent | 9be7dbebfe3614e6dc02f7ea0a13e98fc80fb0f3 (diff) |
Note per Russ on compromise-driven rollover case.
svn path=/docs/repository-structure.txt; revision=569
Diffstat (limited to 'docs/repository-structure.txt')
-rw-r--r-- | docs/repository-structure.txt | 14 |
1 files changed, 12 insertions, 2 deletions
diff --git a/docs/repository-structure.txt b/docs/repository-structure.txt index 9f0a562b..6b0d6a26 100644 --- a/docs/repository-structure.txt +++ b/docs/repository-structure.txt @@ -83,5 +83,15 @@ For those unfamiliar with the notation (borrowed from Lisp): g0001 etc are just "gensym" symbols, ie, the output of some function whose sole purpose is to generate meaningless symbols. -See repository-structure.pdf (source in repository-structure.dot) for -an illustration of the problem and solution. +See images/repository-structure.pdf for an illustration of the problem +and solution. + +There may be compromise-driven rollover cases in which we will need to +reissue all of the children of a node whose key has been compromised. +Whether or not this is necessary depends on whether the master copy of +the authoritative data is safe somewhere else; if it is, and the +resource certificates are just a signed representation of an +authoritative database that has not been compromised, reissuing all of +the descendants may not be necessary, but if the resource certificates +-are- the database, and one level in it has been compromised, it's +probably advisable to reissue all the descendants. |