I am attempting to understand SNMP (in general, and v3). The goal is to include an snmp agent in an embedded device running an RTOS.
I've already been through over a dozen RFCs with at least another dozen more to go. Each one creates more questions than it answers. (1052, 1065, 1067, 1155, 1156, 1157, 1212, 1213, 1592, 1905, 2578, 2579, 2580, 3410, 3411, 3412, 3413, 3414, 3415, 3416, 3417, 3418, 3584... )
I implemented mDNS-SD and 802.1X EAPOL with just a couple RFCs and it wasn't this confusing.
Many of the reviews of books I considered all complain of the same inconsistent and vagueness of the material. I bought a couple books that had better reviews.
Searching online isn't getting anywhere largely because the keywords aren't finding things I want answers to. So I must not even know the best keywords to search with.
Eventually, I decided to just try to reverse engineer what's going on, I installed WireShark on a Linux PC, and the snmpd and snmp tools, so I could sniff it. Here is what I have, and can't align what I see with what I read.
This is a v3 sniff, It's a reply to the first request from a manager. This question is just zeroing in on one of the things that I want to understand. I can't decode and examine a plaintext PDU, because I can't get a request in v2 or v1.
Wireshark shows this reply to a manager. It's apparently the first step in whatever authentication it to be used.
The book I have shows this as the protocol on the wire. And I am trying to parse out the variable bindings.
Here are the variable bindings from Wireshark
A "sequence" that is 15 bytes long (x30 x0f)
This, from the RFC, says that the list is a SEQUENCE of VarBinds, where each VarBind is the object name, and the value in ObjectSyntax. So it's looking okay so far.
Here is the next segment inside the SEQUENCE (Wireshark highlighted all 14 bytes)
An object ID that is 10 bytes long (x06, x0a)
Here is the actual object:
The objectName is the object ID, and it is x2b x6 x1 x6 x3 xf x1 x1 xx4 x0 or (1.3).6.1.6.3.15.1.1.4.0
Given that this is ISO, ORG, DOD, INTERNET, 6?... I have to assume "6" is an object under internet branch I've not yet come across. Likely something to do with the v3 security.
Next, is the value.
This is a type x41 (65), with a length of 1, and a value of 7.
Well, in "ObjectSyntax" what is x41? I can't find it defined anywhere.
For that matter, all these RFCs use words for identifiers, and I can find only a fraction of what their actual numeric values are.
Wireshark knew what it was... It's saying "Counter32"... is that what x41 is supposed to be? If so, it's nowhere near 32 bits. It's only one byte. Again, I'd like to find it's definition.
Also, somewhere, (I can't even recall which RFC) it said the reply to an OID request is to append the value to the requested object, not replace the zero (example: request: 1.3.6.1.4.300.1 -> reply 1.3.6.1.4.300.1.15 so it is a value of 15 ). This OID has a trailing zero, nad I'm not sure why.
Can anyone point me to some useful, concise, condensed information explaining this material? Every RFC requires that I go back and read some previous (and sometimes obsoleted) RFC, and I've now got over 25 of them already. I don't think it should take this many RFCs to be able to write an "simple" snmp agent. A month of researching, and most of what I have to show for it is how to read MIB files. Although that take some mental gymnastics too.
"Simple" is rather deceptive (as more than one book reviewer has stated).
RFC 1157 specifies that SNMP messages are encoded with "a subset of the basic encoding rules of ASN.1". I don't think the official basic encoding rules (BER) specification is available for free, but it's not hard to find explainers online (here's one I found with a simple search). To your question about the 0x41 byte, this is a BER identifier. The 2 most-significant bits (01) tell you the "class" (i.e. something like a namespace) is "application". The "form" bit (0) tells you that it's a primitive type (i.e. not a sequence). Finally the "tag" is 1. Consulting the SNMPv2-SMI MIB (RFC 2578) you can find this definition:
Counter32 ::=
[APPLICATION 1]
IMPLICIT INTEGER (0..4294967295)
You also asked about why a 32-bit integer is encoded with a single byte. This requires you to distinguish between the scope of the SNMP standard versus the ASN.1 standard. ASN.1 only has a single INTEGER type, which 1) has an unlimited range, 2) is always signed (two's complement), and 3) should be encoded in the least number of octets possible. This actually means that a Counter32 (or any other 32-bit unsigned integer type) might use up to 5 bytes for its encoding (see this answer I gave to a question about that).
Finally, you asked about the way the replies are modifying the requested OID. I was confused about this for a long time, but when I figured it out, I realized it's actually pretty simple. I think the best place to start is with this excerpt from RFC 1157:
Each instance of any object type defined in the MIB is identified in
SNMP operations by a unique name called its "variable name." In
general, the name of an SNMP variable is an OBJECT IDENTIFIER of the
form x.y, where x is the name of a non-aggregate object type defined
in the MIB and y is an OBJECT IDENTIFIER fragment that, in a way
specific to the named object type, identifies the desired instance.
This naming strategy admits the fullest exploitation of the semantics
of the GetNextRequest-PDU (see Section 4), because it assigns names
for related variables so as to be contiguous in the lexicographical
ordering of all variable names known in the MIB.
The type-specific naming of object instances is defined below for a
number of classes of object types. Instances of an object type to
which none of the following naming conventions are applicable are
named by OBJECT IDENTIFIERs of the form x.0, where x is the name of
said object type in the MIB definition.
For example, suppose one wanted to identify an instance of the
variable sysDescr The object class for sysDescr is:
iso org dod internet mgmt mib system sysDescr
1 3 6 1 2 1 1 1
Hence, the object type, x, would be 1.3.6.1.2.1.1.1 to which is
appended an instance sub-identifier of 0. That is, 1.3.6.1.2.1.1.1.0
identifies the one and only instance of sysDescr.
So, to summarize, the OID that comes from the MIB doesn't refer to a concrete object, but to the "object type". Each concrete object (i.e. "instance") is identified by a suffix of one or more sub-identifiers (i.e. the y in this explanation). For singleton objects, this suffix is always 0. However, I think most SNMP objects are found in tables, not in singleton objects. I don't actually know of a good explanation of this in the standards, so I'll give it my best shot.
Like any table, SNMP tables are made up of rows and columns. In SNMP, however, the rows are called "entries", and each entry defines a custom type to describe the columns. Here's a simple example from the IF-MIB:
ifTable OBJECT-TYPE
SYNTAX SEQUENCE OF IfEntry
MAX-ACCESS not-accessible
STATUS current
DESCRIPTION
"A list of interface entries. The number of entries is
given by the value of ifNumber."
::= { interfaces 2 }
ifEntry OBJECT-TYPE
SYNTAX IfEntry
MAX-ACCESS not-accessible
STATUS current
DESCRIPTION
"An entry containing management information applicable to a
particular interface."
INDEX { ifIndex }
::= { ifTable 1 }
IfEntry ::=
SEQUENCE {
ifIndex InterfaceIndex,
ifDescr DisplayString,
ifType IANAifType,
ifMtu Integer32,
ifSpeed Gauge32,
ifPhysAddress PhysAddress,
ifAdminStatus INTEGER,
ifOperStatus INTEGER,
ifLastChange TimeTicks,
ifInOctets Counter32,
ifInUcastPkts Counter32,
ifInNUcastPkts Counter32, -- deprecated
ifInDiscards Counter32,
ifInErrors Counter32,
ifInUnknownProtos Counter32,
ifOutOctets Counter32,
ifOutUcastPkts Counter32,
ifOutNUcastPkts Counter32, -- deprecated
ifOutDiscards Counter32,
ifOutErrors Counter32,
ifOutQLen Gauge32, -- deprecated
ifSpecific OBJECT IDENTIFIER -- deprecated
}
So, ifTable has an OID of 1.3.6.1.2.1.2.2, and ifEntry has an OID of 1.3.6.1.2.1.2.2.1. Each item in IfEntry also has its own definition, which includes the OID relative to ifEntry. Generally they match up with the entry's data type, so, for example, ifIndex, as the first column in IfEntry, has an OID of ifEntry.1. Confusingly, when you do a simple Get-Next walk, you will traverse in column-major order, meaning you will get all the ifIndexes, followed by all the ifDescrs, and so on.
So, with all that explained, I'm now prepared to explain the instance identifiers for these tables. Notice above that ifEntry defines
INDEX { ifIndex }
This means, first, that each row is guaranteed to have a unique ifIndex, and, more importantly, that the ifIndex is used as the instance identifier for the entire entry. For example, you can pick any column in the IfEntry data type, let's say ifOperStatus (1.3.6.1.2.1.2.2.1.8), and use Get-Next to find the first instance of that column. Let's say its OID is 1.3.6.1.2.1.2.2.1.8.1, and it's value is 1 (up). The last sub-identifier tells you that it belongs to the row whose ifIndex is 1. To find the name of that interface, you can then query ifDescr.1, and to find its speed setting, you can query ifSpeed.1, and so forth. In this case, it is possible to query ifIndex.1, which will just return 1, but in many tables, the INDEX columns are not-accessible, meaning you can only find out what instances there are by walking some other column. Some tables also use multiple indices, or use OCTET STRING or even OBJECT IDENTIFIER rather than INTEGER typed indices. The rules for encoding and decoding those are in RFC 2578 section 7.7.
I'm working with the IF-MIB and trying to get the text value of an integer using the tools available in net-snmp.
$ snmptranslate -Td IF-MIB::ifAdminStatus.1
IF-MIB::ifAdminStatus.1
ifAdminStatus OBJECT-TYPE
-- FROM IF-MIB
SYNTAX INTEGER {up(1), down(2), testing(3)}
MAX-ACCESS read-write
STATUS current
DESCRIPTION "The desired state of the interface. The testing(3) state
indicates that no operational packets can be passed. When a
managed system initializes, all interfaces start with
ifAdminStatus in the down(2) state. As a result of either
explicit management action or per configuration information
retained by the managed system, ifAdminStatus is then
changed to either the up(1) or testing(3) states (or remains
in the down(2) state)."
::= { iso(1) org(3) dod(6) internet(1) mgmt(2) mib-2(1) interfaces(2) ifTable(2) ifEntry(1) ifAdminStatus(7) 1 }
The SYNTAX here shows that ifAdminStatus has three possible values, up(1), down(2), testing(3). Is it possible to get the string representations (up, down testing) returned for given integer using the net-snmp tools like snmptranslate somehow?
When creating a PKCS#7 signed message with Win32 low level functions like CryptMsgOpenToEncode and CryptMsgUpdate, the resulting message is a message with OID 1.2.840.113549.1.7.2 signedData (PKCS #7), which contains a sequence with OID 1.2.840.113549.1.7.1 data (PKCS #7).
Can I use the low level message functions to change this latter OID? For example, Authenticode uses OID 1.3.6.1.4.1.311.2.1.4 spcIndirectDataContext (Microsoft code signing).
I saw CryptMsgOpenToEncode CMSG_BARE_CONTENT_FLAG flag, but I'm not sure if this is what I want or how to use it.
The (inner) content type of the message is the 5th parameter to CryptMsgOpenToEncode (pszInnerContentObjID).
It should accept any ASCII dotted decimal OID value as input, including the predefined value for the OID you mentioned (SPC_INDIRECT_DATA_OBJID / "1.3.6.1.4.1.311.2.1.4").
I am working on the Agent side of SNMP, and I have a field that is a DateAndTime, which has the syntax:
DateAndTime (OCTET STRING) (SIZE (8 |11)). Hint: 2d-1d-1d,1d:1d:1d.1d,1a1d:1d
The varBind in question returns the timestamp of a certain type of error, which (hopefully) will never happen.
So my question is what the agent should return if the event has not occurred. If I return an empty string, the constraints technically aren't met.
I am not an SNMP expert, so maybe this acceptable?
Changing the MIB to a different definition/Type is a possibility, if that is what I need to do.
The agent should return an error for the incoming requests, by setting error-status to a suitable value (genErr for example), and also set error-index to the proper index in the varbind.
PDU ::= SEQUENCE {
request-id INTEGER (-214783648..214783647),
error-status -- sometimes ignored
INTEGER {
noError(0),
tooBig(1),
noSuchName(2), -- for proxy compatibility
badValue(3), -- for proxy compatibility
readOnly(4), -- for proxy compatibility
genErr(5),
noAccess(6),
wrongType(7),
wrongLength(8),
wrongEncoding(9),
wrongValue(10),
noCreation(11),
inconsistentValue(12),
resourceUnavailable(13),
commitFailed(14),
undoFailed(15),
authorizationError(16),
notWritable(17),
inconsistentName(18)
},
error-index -- sometimes ignored
INTEGER (0..max-bindings),
variable-bindings -- values are sometimes ignored
VarBindList
}
The MIB should define EXACTLY what happens in all circumstances. It is the contract between the agent and manager. If it does not, it is a faulty MIB. If it is a MIB that someone is developing, it needs to be fixed. If it is a public MIB, tell us which, so we can check.
Using net-snmp, table code generated by mib2c -c mib2c.iterate.conf fooBarTable and then heavily hacked.
Unfortunately the table is defined with an Entry of 2 instead of the normal 1. (I didn't do this, I'm trying to make this fit into an existing situation.) The MIB looks something like this:
fooBarTable OBJECT-TYPE
SYNTAX SEQUENCE OF FooBarEntry
MAX-ACCESS not-accessible
STATUS current
DESCRIPTION "blah"
::= { fooMIBObjects 8 }
fooBarEntry OBJECT-TYPE
SYNTAX FooBarEntry
MAX-ACCESS not-accessible
STATUS current
DESCRIPTION "Stuff."
INDEX { ifIndex }
::= { fooBarTable 2 }
When you register the table with net-snmp, you just give it an OID like "...,1,8" (i.e. up to fooBarTable, but not including the Entry). Net-snmp implicitly tacks the .1 to the table OID and then columns, indices, etc.
Is there a semi-supported way to force that entry value to 2? (I.e. without resorting to hacking the bits out of the objects that are passed in to the handler.)
No, sorry: there is no supported way to do that. In part because the MIB you're staring at is not legal under SMIv2.
To implement it, you'd either need to change multiple spots in the agent/helper directory (starting near line 328 of table.c and probably other places) or implement a table entirely from scratch without using the helper modules at all.
But nothing mib2c gives you will solve this for you.