Right now it's just the bare
necessities, e.g. authentication errors don't increment an error
count and are silently dropped. I plan on implementing proper
handling together with the USM-MIB. The security model is
all-or-nothing-read-only. I guess changing it to write access
would make sense, but given that I grant any user access if
LWIP_SNMPV3_GET_USER() is undefined I felt like RO was the safer
option. Though it's only temporary, and I will implement proper
read and write permissions for individual users. What I
additionally plan on implementing is engine ID discovery, which is
optional in the standard. At the moment the implementation only
responds if you send a matching ID. And it defaults to the test ID
"000000000000000000000002" from RFC3414, which allows for
convenient crypto testing. I've also left some TODO-comments in
areas that I want to reconsider again later. Regarding maintenance
I'd be willing to help out, after all it's in my own interest.
That said, it has to stay within reasonable bounds. My job
requires me to travel a lot and keeps me quite busy.
-Elias
On 24/03/16 22:13, Dirk Ziegelmeier wrote:
Thanks Elias!!!
I'll have a
look at it, but it may take some time because I also want my
colleague who (re-)wrote much of the agent (Martin Hentschel)
have a look at it.
|