[EMAIL PROTECTED] schrieb:
Is it possible to sign the same set of messages with multiple
algorithms (by sending multiple Payload and Signature Blocks) to
provide smooth algorithm transition?
I think it should be possible.
A message can belong to multiple signature groups, so an originator
could e.g. use one signature group with SG="0" and VER="0111" and a
second one with SG="0" and VER="0121".
This leads me to a problem in 4.2.3.: "In all cases, no more than 192
distinct Signature Groups (0-191) are permitted."
IMHO that should be removed or changed to "In all cases, no more than
192 distinct SPRI values (0-191) are permitted."
Such a limit would clearly prevent using parallel versions for SG="1".
Or in another scenario: my syslogd uses SG="3" to assign one signature
group per destination. So that limit would prevent the use of more than
192 destinations.
Section 4.2.5: When counting messages for the "First Message Number"
field, are Signature Blocks and Certificate Blocks also counted?
Whether the FMN counts syslog-sign messages depends on whether these are
hashed and included in Signature Blocks.
Should earlier Signature Block and/or Certificate Block messages
be included in Hash Blocks?
4.1:
The syslog messages defined as part of syslog-sign themselves
(Signature Block messages and Certificate Block messages) do not need
to be signed by a Signature Block. Collectors that implement syslog-
sign know to distinguish syslog messages that are associated with
syslog-sign from those that are subjected to signing and process them
differently.
I guess there are arguments for both positions but IMHO syslog-sign
messages should not be hashed and included because
a) I see syslog-sign as a kind of session-layer that should disappear
after the verification is done and some log-analyzer (=application
layer) takes over, and
b) the verifier will have to know to treat them differently anyway and
leaving them out avoids several problems and special cases.
Either way the standard has to decide for one position to ensure
interoperability. Otherwise one might have a sender that does include
hashes of syslog-sign messages in its Signature Blocks and a verifier
that will not check them but report lost messages.
--
Martin
_______________________________________________
Syslog mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/syslog
