pief / python-netsnmpagent Goto Github PK

View Code? Open in Web Editor NEW

67.0 67.0 38.0 2.3 MB

Allows to write net-snmp subagents in Python.

License: GNU Lesser General Public License v3.0

Python 95.54% Makefile 4.46%

python-netsnmpagent's People

Contributors

Stargazers

Watchers

python-netsnmpagent's Issues

Can't reach subagent from system-wide level

Hello.

First of all, thank you for that library, since it appears to be the most alive Pyhon-SNMP implementation.

I tried to run simple agent example, from the master branch, that was successful. After that I tried to do as comments say at the beginning of simple_agent.py.

However this time I had response:

snmpwalk -v 2c -c secret -M+. localhost SIMPLE-MIB::simpleMIB
SIMPLE-MIB::simpleMIB = No Such Object available on this agent at this OID

Also, I noted that nowadays <rosecret> is called simply secret inside the /etc/snmp/snmpd.conf.

And in comments:

Then, from a separate console and from inside the python-netsnmpagent
directory, you can run

I believe you meant from /examples directory, since SIMPLE_MIB.txt is available only there?

Updated DisplayString can not be longer than initval

I am creating a displayString with code like the following. It seems the updated value for a display string gets cuts off at the initval length of the string. If I make the initval longer or shorter, the updated value reflect this as well.

version = agent.DisplayString(oidstr="PRODUCT-MIB::version", initval="123")
version.update("98765")
>> Result from SNMP PRODUCT-MIB::version is: 987

Separate the update process from request processing

Hi,

First of all, I wanted to thank you. It's really hard to find AgentX implementations written in Python that actually work.

But I have to admit I'm having trouble with the update process.

In your example, you launch the updates when agent.check_and_process() unlocks (when receiving a request).

However in my case, I need the data to be ready sooner. I can't afford to lose time to get the new values.

I am thinking about using another thread, which will deal with the updates, but it implies to use locks almost everywhere, and I want to avoid
that as much as possible.

Do you know if it is possible to launch the updates BEFORE (and independently from) a request reception?

Thank you in advance for your answer.

Best regards,
Salem

Couldn't install from github

We need the latest version of netsnmpagent package with proper python3 support, but unfortunately couldn't install it directly from github since there is no setup.py and version in pypi archive doesn't work with python3.4.

Please make project installable directly from github or publish new fresh release on pypi.

pkg_resources.DistributionNotFound exception when using netsnmpagent.py directly from source tree

c77f52a introduced code that uses pkg_resources to get the module's version, as defined in setup.py. When using netsnmpagent.py directly from the source tree, eg. by running the example agents, this fails unless a "make install" is performed:

[...]
* Starting the simple example agent...
Traceback (most recent call last):
  File "simple_agent.py", line 47, in <module>
    import netsnmpagent
  File "/home/pief/python-netsnmpagent/netsnmpagent.py", line 29, in <module>
    __version__ = pkg_resources.require("netsnmpagent")[0].version
  File "/usr/lib/python2.7/site-packages/pkg_resources.py", line 690, in require
    needed = self.resolve(parse_requirements(requirements))
  File "/usr/lib/python2.7/site-packages/pkg_resources.py", line 588, in resolve
    raise DistributionNotFound(req)
pkg_resources.DistributionNotFound: netsnmpagent
* Cleaning up...

getRegistered(context) throws exception when OctetString is DateTime

I'm setting an OctetString as follows as an SNMP DateTime object

I'm not sure if the way I'm setting the DateTime is correct for SNMP, but when I do a snmpwalk, I'm able to see the variable just fine. Except the program blows up when I call dump_registered()

setup

import struct
import time

# setup the agent as in simple_agent.py example, then

time_millisecs = 1577836800000

# convert milliseconds to an OctetString
value = time.gmtime(time_millisecs / 1000)
value = struct.pack(
                            ">HBBBBBB",
                            value.tm_year,
                            value.tm_mon,
                            value.tm_mday,
                            value.tm_hour,
                            value.tm_min,
                            value.tm_sec,
                            0,
                        )
new_row.setRowCell(2, agent.OctetString(value))

# the above is fine, but it then blows up when the following is called
dump_registered(agent)

stacktrace

Traceback (most recent call last):
  File "./snmp_agent.py", line 465, in <module>
    main()
  File "./snmp_agent.py", line 435, in main
    dump_registered(agent)
  File "./snmp_agent.py", line 342, in dump_registered
    variables = agent.getRegistered(context)
  File "~/my_venv/lib/python3.8/site-packages/netsnmpagent.py", line 949, in getRegistered
    "value": snmpobj.value()
  File "~/my_venv/lib/python3.8/site-packages/netsnmpagent.py", line 895, in value
    retdict[indices][int(data.contents.column)] = u(ctypes.string_at(data.contents.data.string, data.contents.data_len))
  File "~/my_venv/lib/python3.8/site-packages/netsnmpagent.py", line 51, in u
    return s if isinstance("Test", bytes) else s.decode(locale.getpreferredencoding())
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xe4 in position 1: invalid continuation byte
Exception ignored in: <module 'threading' from '/usr/lib/python3.8/threading.py'>
Traceback (most recent call last):
  File "/usr/lib/python3.8/threading.py", line 1388, in _shutdown
    lock.acquire()
KeyboardInterrupt: 
* Cleaning up...

Can't use IpAddress as table index

Using an IpAddress object as table index fails with

AttributeError: 'IpAddress' object has no attribute '_asntype'

Example snippet for how to reproduce:

tableWithIpIndex = agent.Table(
    oidstr = "SIMPLE-MIB::thirdTable",
    indexes = [
        agent.IpAddress()
    ],
    columns = [
        (2, agent.DisplayString("")),
        (3, agent.IpAddress("0.0.0.0"))
    ],
)
row1 = tableWithIpIndex.addRow([agent.IpAddress("192.168.0.1")])
row1.setRowCell(2, agent.DisplayString("Some value"))
row1.setRowCell(3, agent.IpAddress("192.168.0.1"))

The first part of a potential fix is simple, just initialize IpAddress with the missing attributes, e.g.:

class IpAddress(object):
    def __init__(self):
        self._flags = WATCHER_FIXED_SIZE
        self._asntype = ASN_IPADDRESS
        self._cvar = ctypes.c_uint(0)
        self._data_size = ctypes.sizeof(self._cvar)
        self._max_size  = self._data_size
        self.update(initval)

Now comes the tricky part. There's an open Net-SNMP Bug 2136. Its impact is that Net-SNMP expects an IP address in network byte order if used as column instance, but in host byte order if used as index. This hits us if we're on little-endian machines like Intel x86-64.

Without further measures our example would give the correct cell value, but a reversed index

AGENTX-MIB::agentxObjects.30187.1.2.6.1.3.1.0.168.192 = IpAddress: 192.168.0.1

I'll send a pull request soon, with a suggestion how to fix the second issue too. It will produce expected output (at least if RFC 3416, section 4.2.2.1 is taken as a reference for how it should be)

AGENTX-MIB::agentxObjects.30187.1.2.6.1.3.192.168.0.1 = IpAddress: 192.168.0.1

python: undefined symbol: snmp_register_callback

Hello,

I am trying to use the Python package netsnmpagent-0.5.3.tar.gz to setup a sub-agent. When I run the example "run_simple_agent.sh" I get the following error:

AttributeError: python: undefined symbol: snmp_register_callback

Script output:

Preparing snmpd environment...
Our snmpd instance is now listening on localhost, port 5555.
From a second console, use the net-snmp command line utilities like this:

cd /home/robm/snmp/src/python/netsnmpagent-0.5.3/examples
snmpwalk -v 2c -c public -M+. localhost:5555 SIMPLE-MIB::simpleMIB
snmptable -v 2c -c public -M+. -Ci localhost:5555 SIMPLE-MIB::firstTable
snmpget -v 2c -c public -M+. localhost:5555 SIMPLE-MIB::simpleInteger.0
snmpset -v 2c -c simple -M+. localhost:5555 SIMPLE-MIB::simpleInteger.0 i 123
Starting the simple example agent...
Traceback (most recent call last):
File "simple_agent.py", line 47, in
import netsnmpagent
File "/home/robm/snmp/src/python/netsnmpagent-0.5.3/netsnmpagent.py", line 24, in
from netsnmpapi import *
File "/home/robm/snmp/src/python/netsnmpagent-0.5.3/netsnmpapi.py", line 55, in
for f in [ libnsa.snmp_register_callback ]:
File "/usr/lib/python2.7/ctypes/init.py", line 378, in getattr
func = self.getitem(name)
File "/usr/lib/python2.7/ctypes/init.py", line 383, in getitem
func = self._FuncPtr((name_or_ordinal, self))
AttributeError: python: undefined symbol: snmp_register_callback

I am using the net-snmp-5.7.3 package which I configured the following way:
configure --with-mib-modules=smux

Any help will be appreciated.

Kind regards,
Rob

Callback support?

Thank you for doing this work! I experienced the same problem as you -- hardly any support for SNMP agents other than through C or Perl.
I think your model is nicer than the "push persist" mode of snmpd, but snmpd is still better suited for interactions with the environment -- local modifications over SNMP, and on-the-spot lookups in response to SNMP inquiries.
Have you considered permitting callback functions to make that sort of thing possible with your agent? Matching against a beginning-of-OID as in snmpd should work nicely. Such a construct could implement more dynamic getting, as wel as setting of values than now possible with python-netsnmpagent.

I can't update row Counter64

Hi,

I ran the simple agent to read linux network interfaces eth0, eth1 and eth2:

{'NETWORKZ-MIB::firstTable': {'type': 'Table',
'value': {0: {2: {'type': 'OctetString', 'value': 'Unknown place'},
3: {'type': 'Counter64', 'value': 0}},
'1': {2: 'eth0', 3: 204064872L},
'2': {2: 'eth1', 3: 134064843},
'3': {2: 'eth2', 3: 988190442}}},

When I execute:

snmpwalk -v 2c -c public -M+. localhost:5555 NETWORKZ-MIB::simpleMIB
shows me:
NETWORKZ-MIB::ifDescr."1" = STRING: eth0
NETWORKZ-MIB::ifDescr."2" = STRING: eth1
NETWORKZ-MIB::ifDescr."3" = STRING: eth2
NETWORKZ-MIB::ifInBytes."1" = Counter64: 204064872
NETWORKZ-MIB::ifInBytes."2" = Counter64: 134064843
NETWORKZ-MIB::ifInBytes."3" = Counter64: 988190442

When I execute again the snmpwalk, the result is the same. I can't update Counter64 values

The code is:

Add the first table row

firstTableRow1 = firstTable.addRow([agent.DisplayString("1")])
firstTableRow1.setRowCell(2, agent.DisplayString("eth0"))
firstTableRow1.setRowCell(3, agent.Counter64(get_in_bytes("eth0")))

Add the second table row

firstTableRow2 = firstTable.addRow([agent.DisplayString("2")])
firstTableRow2.setRowCell(2, agent.DisplayString("eth1"))
#firstTableRow2.setRowCell(3, agent.Integer32(28))
firstTableRow2.setRowCell(3, agent.Counter64(get_in_bytes("eth1")))

Add the third table row

firstTableRow3 = firstTable.addRow([agent.DisplayString("3")])
firstTableRow3.setRowCell(2, agent.DisplayString("eth2"))
#firstTableRow3.setRowCell(3, agent.Integer32(18))
firstTableRow3.setRowCell(3, agent.Counter64(get_in_bytes("eth2")))

The function get_in_bytes() allow me to get the input bytes on the interface

Juan Carlos

Segfault with Table.clear()

I am using an agent that builds a table and periodically clears it. Rarely I receive a segmentation fault. Narrowing down the code and using valgrind, it spots invalid memory reads.

The attached script demonstrates:

$ valgrind python ./snmp-agent.py 
==169269== Memcheck, a memory error detector
==169269== Copyright (C) 2002-2022, and GNU GPL'd, by Julian Seward et al.
==169269== Using Valgrind-3.19.0 and LibVEX; rerun with -h for copyright info
==169269== Command: python ./snmp-agent.py
==169269== 
[Error] MIB search path: ...
[Error] Cannot find module (UCD-SNMP-MIB): At line 1 in (none)
[Error] Cannot find module (UCD-DEMO-MIB): At line 1 in (none)
[Error] Cannot find module (NET-SNMP-AGENT-MIB): At line 1 in (none)
[Error] Cannot find module (NET-SNMP-PASS-MIB): At line 1 in (none)
[Error] Cannot find module (NET-SNMP-EXTEND-MIB): At line 1 in (none)
[Error] Cannot find module (UCD-DLMOD-MIB): At line 1 in (none)
[Error] Cannot find module (NET-SNMP-VACM-MIB): At line 1 in (none)
Cycle 0
Cycle 1
==169269== Invalid read of size 8
==169269==    at 0x56DC16E: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x576FF4: ??? (in /usr/bin/python3.11)
==169269==    by 0x53AC2B: PyObject_Vectorcall (in /usr/bin/python3.11)
==169269==    by 0x52B93F: _PyEval_EvalFrameDefault (in /usr/bin/python3.11)
==169269==    by 0x52360A: PyEval_EvalCode (in /usr/bin/python3.11)
==169269==    by 0x647496: ??? (in /usr/bin/python3.11)
==169269==    by 0x644D4E: ??? (in /usr/bin/python3.11)
==169269==    by 0x65100F: ??? (in /usr/bin/python3.11)
==169269==    by 0x650D5A: _PyRun_SimpleFileObject (in /usr/bin/python3.11)
==169269==    by 0x650B83: _PyRun_AnyFileObject (in /usr/bin/python3.11)
==169269==    by 0x64F90E: Py_RunMain (in /usr/bin/python3.11)
==169269==    by 0x6275C6: Py_BytesMain (in /usr/bin/python3.11)
==169269==  Address 0x6616ee0 is 32 bytes inside a block of size 48 free'd
==169269==    at 0x484317B: free (vg_replace_malloc.c:872)
==169269==    by 0x588BC04: netsnmp_table_data_delete_row (table_data.c:144)
==169269==    by 0x588CCB2: netsnmp_table_dataset_remove_and_delete_row (table_dataset.c:201)
==169269==    by 0x588CCB2: netsnmp_table_dataset_remove_and_delete_row (table_dataset.c:192)
==169269==    by 0x4862F79: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x486240D: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x4862B0C: ffi_call (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x56D942A: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x56D8612: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x517EB2: _PyObject_MakeTpCall (in /usr/bin/python3.11)
==169269==    by 0x52B93F: _PyEval_EvalFrameDefault (in /usr/bin/python3.11)
==169269==    by 0x52360A: PyEval_EvalCode (in /usr/bin/python3.11)
==169269==    by 0x647496: ??? (in /usr/bin/python3.11)
==169269==  Block was alloc'd at
==169269==    at 0x48455EF: calloc (vg_replace_malloc.c:1328)
==169269==    by 0x588D7AD: netsnmp_table_data_set_create_row_from_defaults (table_dataset.c:292)
==169269==    by 0x4862F79: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x486240D: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x4862B0C: ffi_call (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x56D942A: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x56D8612: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x517EB2: _PyObject_MakeTpCall (in /usr/bin/python3.11)
==169269==    by 0x52B93F: _PyEval_EvalFrameDefault (in /usr/bin/python3.11)
==169269==    by 0x51F066: _PyObject_FastCallDictTstate (in /usr/bin/python3.11)
==169269==    by 0x565972: ??? (in /usr/bin/python3.11)
==169269==    by 0x517E54: _PyObject_MakeTpCall (in /usr/bin/python3.11)
==169269== 
==169269== Invalid read of size 8
==169269==    at 0x56DC4FB: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x543E9B: _PyObject_GenericGetAttrWithDict (in /usr/bin/python3.11)
==169269==    by 0x51E473: PyObject_GetAttr (in /usr/bin/python3.11)
==169269==    by 0x52BF94: _PyEval_EvalFrameDefault (in /usr/bin/python3.11)
==169269==    by 0x52360A: PyEval_EvalCode (in /usr/bin/python3.11)
==169269==    by 0x647496: ??? (in /usr/bin/python3.11)
==169269==    by 0x644D4E: ??? (in /usr/bin/python3.11)
==169269==    by 0x65100F: ??? (in /usr/bin/python3.11)
==169269==    by 0x650D5A: _PyRun_SimpleFileObject (in /usr/bin/python3.11)
==169269==    by 0x650B83: _PyRun_AnyFileObject (in /usr/bin/python3.11)
==169269==    by 0x64F90E: Py_RunMain (in /usr/bin/python3.11)
==169269==    by 0x6275C6: Py_BytesMain (in /usr/bin/python3.11)
==169269==  Address 0x6616ee0 is 32 bytes inside a block of size 48 free'd
==169269==    at 0x484317B: free (vg_replace_malloc.c:872)
==169269==    by 0x588BC04: netsnmp_table_data_delete_row (table_data.c:144)
==169269==    by 0x588CCB2: netsnmp_table_dataset_remove_and_delete_row (table_dataset.c:201)
==169269==    by 0x588CCB2: netsnmp_table_dataset_remove_and_delete_row (table_dataset.c:192)
==169269==    by 0x4862F79: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x486240D: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x4862B0C: ffi_call (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x56D942A: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x56D8612: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x517EB2: _PyObject_MakeTpCall (in /usr/bin/python3.11)
==169269==    by 0x52B93F: _PyEval_EvalFrameDefault (in /usr/bin/python3.11)
==169269==    by 0x52360A: PyEval_EvalCode (in /usr/bin/python3.11)
==169269==    by 0x647496: ??? (in /usr/bin/python3.11)
==169269==  Block was alloc'd at
==169269==    at 0x48455EF: calloc (vg_replace_malloc.c:1328)
==169269==    by 0x588D7AD: netsnmp_table_data_set_create_row_from_defaults (table_dataset.c:292)
==169269==    by 0x4862F79: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x486240D: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x4862B0C: ffi_call (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x56D942A: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x56D8612: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x517EB2: _PyObject_MakeTpCall (in /usr/bin/python3.11)
==169269==    by 0x52B93F: _PyEval_EvalFrameDefault (in /usr/bin/python3.11)
==169269==    by 0x51F066: _PyObject_FastCallDictTstate (in /usr/bin/python3.11)
==169269==    by 0x565972: ??? (in /usr/bin/python3.11)
==169269==    by 0x517E54: _PyObject_MakeTpCall (in /usr/bin/python3.11)
==169269== 
==169269== Invalid read of size 8
==169269==    at 0x56E4DC9: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x56E2949: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x56D8612: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x517EB2: _PyObject_MakeTpCall (in /usr/bin/python3.11)
==169269==    by 0x52B93F: _PyEval_EvalFrameDefault (in /usr/bin/python3.11)
==169269==    by 0x52360A: PyEval_EvalCode (in /usr/bin/python3.11)
==169269==    by 0x647496: ??? (in /usr/bin/python3.11)
==169269==    by 0x644D4E: ??? (in /usr/bin/python3.11)
==169269==    by 0x65100F: ??? (in /usr/bin/python3.11)
==169269==    by 0x650D5A: _PyRun_SimpleFileObject (in /usr/bin/python3.11)
==169269==    by 0x650B83: _PyRun_AnyFileObject (in /usr/bin/python3.11)
==169269==    by 0x64F90E: Py_RunMain (in /usr/bin/python3.11)
==169269==  Address 0x6616ee0 is 32 bytes inside a block of size 48 free'd
==169269==    at 0x484317B: free (vg_replace_malloc.c:872)
==169269==    by 0x588BC04: netsnmp_table_data_delete_row (table_data.c:144)
==169269==    by 0x588CCB2: netsnmp_table_dataset_remove_and_delete_row (table_dataset.c:201)
==169269==    by 0x588CCB2: netsnmp_table_dataset_remove_and_delete_row (table_dataset.c:192)
==169269==    by 0x4862F79: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x486240D: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x4862B0C: ffi_call (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x56D942A: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x56D8612: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x517EB2: _PyObject_MakeTpCall (in /usr/bin/python3.11)
==169269==    by 0x52B93F: _PyEval_EvalFrameDefault (in /usr/bin/python3.11)
==169269==    by 0x52360A: PyEval_EvalCode (in /usr/bin/python3.11)
==169269==    by 0x647496: ??? (in /usr/bin/python3.11)
==169269==  Block was alloc'd at
==169269==    at 0x48455EF: calloc (vg_replace_malloc.c:1328)
==169269==    by 0x588D7AD: netsnmp_table_data_set_create_row_from_defaults (table_dataset.c:292)
==169269==    by 0x4862F79: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x486240D: ??? (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x4862B0C: ffi_call (in /usr/lib/x86_64-linux-gnu/libffi.so.8.1.2)
==169269==    by 0x56D942A: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x56D8612: ??? (in /usr/lib/python3.11/lib-dynload/_ctypes.cpython-311-x86_64-linux-gnu.so)
==169269==    by 0x517EB2: _PyObject_MakeTpCall (in /usr/bin/python3.11)
==169269==    by 0x52B93F: _PyEval_EvalFrameDefault (in /usr/bin/python3.11)
==169269==    by 0x51F066: _PyObject_FastCallDictTstate (in /usr/bin/python3.11)
==169269==    by 0x565972: ??? (in /usr/bin/python3.11)
==169269==    by 0x517E54: _PyObject_MakeTpCall (in /usr/bin/python3.11)
==169269== 
==169269== 
==169269== HEAP SUMMARY:
==169269==     in use at exit: 988,875 bytes in 9,574 blocks
==169269==   total heap usage: 26,267 allocs, 16,693 frees, 14,983,444 bytes allocated
==169269== 
==169269== LEAK SUMMARY:
==169269==    definitely lost: 0 bytes in 0 blocks
==169269==    indirectly lost: 0 bytes in 0 blocks
==169269==      possibly lost: 0 bytes in 0 blocks
==169269==    still reachable: 988,875 bytes in 9,574 blocks
==169269==         suppressed: 0 bytes in 0 blocks
==169269== Rerun with --leak-check=full to see details of leaked memory
==169269== 
==169269== For lists of detected and suppressed errors, rerun with: -s
==169269== ERROR SUMMARY: 4 errors from 3 contexts (suppressed: 0 from 0)

snmp-agent.py.txt

Howewer, if I apply the following simple diff:

--- netsnmpagent.py.orig	2023-12-26 19:04:50.313345452 +0100
+++ netsnmpagent.py	2023-12-26 19:05:59.302824678 +0100
@@ -912,14 +912,11 @@
 				return retdict
 
 			def clear(self):
-				row = self._dataset.contents.table.contents.first_row
-				while bool(row):
-					nextrow = row.contents.next
+				while bool(row := self._dataset.contents.table.contents.first_row):
 					libnsX.netsnmp_table_dataset_remove_and_delete_row(
 						self._dataset,
 						row
 					)
-					row = nextrow
 				if self._counterobj:
 					self._counterobj.update(0)

The mentioned warning from valgrind disappears. And since then, my application is not crashing. Howewer, I am not sure why this does solve my issue, I dont know python's ctype's internals.

No Gauge32 available

Hi,

I want to implement a gauge, but netsnmpAgent has no Gauge32.
e.g.

activeSessionAvg = agent.Gauge32(
    oidstr   = ....,
    initval  = 0,
    writable = False
)

Error message is:

AttributeError: 'netsnmpAgent' object has no attribute 'Gauge32'

Please consider pushing to pypi

Hi,

I'm using this module to run an SNMP subagent, thank you for your work.

It'd be great to have a version of this in pipy that can run the examples.

Thanks!

Edit: I see the tags are there :)

How to update Counter64 objects in the table

I have declared table like this

MyTable = agent.Table(
    oidstr  = "MY-MIB::MyTable",
    indexes = [
        agent.DisplayString(),
        agent.Unsigned32()
    ],
    columns = [
        (1, agent.DisplayString("NA")),
        (2, agent.Unsigned32(1)),
        (3, agent.DisplayString("NA")),
        (4, agent.DisplayString("NA")),
        (5, agent.Unsigned32(0)),
        (6, agent.Unsigned32(0)),
        (7, agent.DisplayString("NA")),
        (8, agent.DisplayString("NA")),
        (9, agent.Integer32(0)),
        (10, agent.Counter64(0)),
        (11, agent.Counter64(0)),
        (12, agent.Counter64(0)),
        (13, agent.Counter64(0)),
        (14, agent.Integer32(0)),
        (15, agent.Integer32(0)),
        (16, agent.Integer32(0)),
    ]
)

but the counter64 object is no updating when call it in setRowCell

rd.setRowCell(13, agent.Counter64(5000))

I am missing something?

Can not update table entry with python code

Hi,

I set up a table with python. Now I want to change values at runtime from inside the python script. Therefor I tried to store references of the types of a row:

If I try...:

rcnt = agent.Counter32(500)
rflag = agent.Integer32(99)

row = self._stats_table.addRow([agent.Integer32(1)])
row.setRowCell(2, agent.DisplayString("Some name"))
row.setRowCell(3, rcnt)
row.setRowCell(4, rflag)

...

rcnt.update(1000)
rflag.update(5000)

...the table is not updated and I do not understand why. I always have to clear() the whole table and set values newly. Is that the only way doing it? Unfortunately the simple-example only demonstrates updates on scalar types.

No support for OctetStrings with zeros.

The library uses ctypes.create_string_buffer and ctypes.string_at etc.
This causes an issue when trying to work with buffers that should be returned as hexadecimal.
For example if I want to return IPv6 address of MAC as an octet string and one of the bytes is zero I will get a partial result (the prefix of the buffer until we hit the zero byte).

This can most likely be fixed by using ctypes.c_ubyte * N to create the buffers instead ctype.create_string_buffer

OSError: /usr/lib64/libnetsnmphelpers.so.10: undefined symbol: snmp_free_var

Hi Pieter-

This is a shot in the dark here, but I'm encountering an issue with loading the netsnmp shared libraries in python, and I thought maybe you had some insight:

>>> ctypes.CDLL("libnetsnmphelpers.so.10")
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib64/python2.6/ctypes/__init__.py", line 353, in __init__
    self._handle = _dlopen(self._name, mode)
OSError: /usr/lib64/libnetsnmphelpers.so.10: undefined symbol: snmp_free_var

Yet that symbol exists in that library:

# readelf -Ws  /usr/lib64/libnetsnmphelpers.so.10 | grep snmp_free_var
    12: 0000000000000000     0 NOTYPE  GLOBAL DEFAULT  UND snmp_free_var
    56: 0000000000000000     0 NOTYPE  GLOBAL DEFAULT  UND snmp_free_varbind

I've tried it with ctypes.RTLD_GLOBAL set to the mode, but that doesn't do anything.

Here is the version of the OS and library:

# cat /etc/redhat-release 
CentOS release 5.8 (Final)
# rpm -q net-snmp
net-snmp-5.3.2.2-17.el5_8.1

Interestingly though, it works on CentOS 6.x

>>> ctypes.cdll.LoadLibrary(ctypes.util.find_library("netsnmphelpers"))
<CDLL 'libnetsnmphelpers.so.20', handle 9d7890 at 7fe1caf76690>

# cat /etc/redhat-release 
CentOS release 6.5 (Final)
# rpm -q net-snmp
net-snmp-5.5-49.el6_5.1.x86_64

Any insight would help, but I understand if this cannot be helped.

Wrong RFC in README

As far as I understand correct RFC for SNMP AgentX protocol is RFC2741 and not RFC2747 as mentioned in README.

Strings inside tables erroneously get a dot (".") suffix

Under openSUSE 13.1, string objects inside tables seem to erroneously get a dot (".") suffix that ordinary string objects don't:

pief@e6400:~/python-netsnmpagent/examples> snmpwalk -v 2c -c public -M+. localhost:5555 SIMPLE-MIB::simpleMIB
[...]
SIMPLE-MIB::simpleOctetString.0 = STRING: Hello World
SIMPLE-MIB::simpleDisplayString.0 = STRING: Nice to meet you
SIMPLE-MIB::firstTableNumber.0 = Gauge32: 3
SIMPLE-MIB::firstTableEntryDesc."aa" = STRING: Prague.
SIMPLE-MIB::firstTableEntryDesc."ab" = STRING: Barcelona.
SIMPLE-MIB::firstTableEntryDesc."bb" = STRING: Unknown place.
[...]

I can't remember seeing this under openSUSE 12.2, so it might be due to different net-snmp versions. Needs investigation.

References to exampleCounter in simple_agent.py

Was just running the simple_agent.py to get my feet wet. There was a minor bug with a few references to example (rather than simple).

exampleCounter64
EXAMPLE-MIB
exampleCounter32Context2
exampleCounter64Context2

Cannot update table entry

Hi,

I'm trying to update a table entry with snmpset but I always get an error.

I used the MIB "SIMPLE-MIB.txt" and the python program "simple_agent.py" to play with. First I changed the lines "MAX-ACCESS" to "read-write". Next I updated a simple object which worked fine and then I tried the update a table entry which failed.

Traces:

 $ snmpget -v 2c -c public -m$PWD/mibs/SIMPLE-MIB.txt localhost:5555 SIMPLE-MIB::simpleInteger.0
SIMPLE-MIB::simpleInteger.0 = INTEGER: 0
$ snmpset -v 2c -c private -m$PWD/mibs/SIMPLE-MIB.txt localhost:5555 SIMPLE-MIB::simpleInteger.0 i 123
SIMPLE-MIB::simpleInteger.0 = INTEGER: 123
$ snmpget -v 2c -c public -m$PWD/mibs/SIMPLE-MIB.txt localhost:5555 SIMPLE-MIB::simpleInteger.0       
SIMPLE-MIB::simpleInteger.0 = INTEGER: 123
$  snmpget -v 2c -c public -m$PWD/mibs/SIMPLE-MIB.txt localhost:5555 SIMPLE-MIB::secondTableEntryDesc.1
SIMPLE-MIB::secondTableEntryDesc.1 = STRING: foo0
$  snmpset -v 2c -c private -m$PWD/mibs/SIMPLE-MIB.txt localhost:5555 SIMPLE-MIB::secondTableEntryDesc.1 s hello
Error in packet.
Reason: notWritable (That object does not support modification)
Failed object: SIMPLE-MIB::secondTableEntryDesc.1

Any ideas?

Kind regards,
Rob

self._status update failed at agent reconnexion

In netsnmpagent.py, line 183, the regex match is not correct :

The line is:
and re.match("AgentX subagent connected", msgtext):

It should be started with .* like this:
and re.match(".*AgentX subagent connected", msgtext):

Because the full message is :
"NET-SNMP version 5.9 AgentX subagent connected"

And the match is never taken, without the .*

Thanks,
Amandine