超过1,000台主机的PYSNMP扭曲客户端出现未处理错误

问题描述

我有以下代码：

from twisted.internet import reactor
from twisted.internet import defer, task
from pysnmp.entity import engine, config
from pysnmp.carrier.twisted import dispatch
from pysnmp.carrier.twisted.dgram import udp
from pysnmp.entity.rfc3413.twisted import cmdgen


import __webimport__
import tools.config
from tools.database import makedsn
import psycopg2


def cmp_varBinds(varBind, varName):
    if varName[0] in str(varBind[0]):
        return True


def cbFun(cbCtx, ip, varNames):
    (errorIndication, errorStatus, errorIndex, varBinds) = cbCtx
    if varBinds and any(map(cmp_varBinds, varBinds[0], varNames)):
        print ip, [str(x[1]) for x in varBinds[0]]
        df = defer.Deferred()
        df.addCallback(cbFun, ip=ip, varNames=varNames)
        return df  # This also indicates that we wish to continue walkin


def parallel(iterable, count, callable, *args, **named):
    coop = task.Cooperator()
    work = (callable(elem, *args, **named) for elem in iterable)
    return defer.DeferredList([coop.coiterate(work) for i in xrange(count)])





def fetch(host):
    id, ip, community, hc = host
    snmpEngine = engine.SnmpEngine()
    snmpEngine.registerTransportDispatcher(dispatch.TwistedDispatcher())
    config.addV1System(snmpEngine, 'test-agent', community)
    config.addTargetParams(snmpEngine, 'myParams', 'test-agent', 'noAuthNoPriv', 1)

    config.addTargetAddr(
                     snmpEngine, 'myRouter', config.snmpUDPDomain,
                     (ip, 161), 'myParams', timeout=1
    )

    # Transport
    config.addSocketTransport(
                          snmpEngine,
                          udp.domainName,
                          udp.UdpTwistedTransport().openClientMode()
    )

    getCmdGen = cmdgen.NextCommandGenerator()
    varNames = [('1.3.6.1.2.1.2.2.1.11', None),
            ('1.3.6.1.2.1.2.2.1.12', None),
            ('1.3.6.1.2.1.2.2.1.13', None),
            ('1.3.6.1.2.1.2.2.1.14', None)]
    df = getCmdGen.sendReq(snmpEngine, 'myRouter', varNames)
    df.addCallback(cbFun, ip=ip, varNames=varNames)
    return df


dsn = makedsn(tools.config.main_db)
connection = psycopg2.connect(dsn)
cursor = connection.cursor()
cursor.execute("""SELECT e.id, e.ip, e.snmpcomm, e.hccnt
           FROM snmp_ports sp, equipment e
           WHERE e.snmp = 'Y' and sp.equipment = e.id
           GROUP BY e.id,e.ip,e.snmpcomm,e.hccnt
           ORDER BY e.id""")
hosts = cursor.fetchall()


finished = parallel(hosts, len(hosts), fetch)
finished.addErrback(log.err)
finished.addCallback(lambda ign: reactor.stop())
reactor.run()

我从数据库中获取了4000台主机，并向每个主机发出请求。如果我在SQL查询中设置了LIMIT 1000，它就可以正常工作。但是当主机数量超过1000台时，我会收到一个错误：

Unhandled Error
Traceback (most recent call last):
  File "crawler.py", line 98, in <module>

  File "/home/kalombo/.virtualenvs/dev/local/lib/python2.7/site-packages/twisted/internet/base.py", line 1192, in run

  File "/home/kalombo/.virtualenvs/dev/local/lib/python2.7/site-packages/twisted/internet/base.py", line 1201, in mainLoop

--- <exception caught here> ---
  File "/home/kalombo/.virtualenvs/dev/local/lib/python2.7/site-packages/twisted/internet/base.py", line 824, in runUntilCurrent

  File "/home/kalombo/.virtualenvs/dev/local/lib/python2.7/site-packages/pysnmp/carrier/base.py", line 52, in _cbFun

  File "/home/kalombo/.virtualenvs/dev/local/lib/python2.7/site-packages/pysnmp/entity/engine.py", line 64, in __receiveMessageCbFun

  File "/home/kalombo/.virtualenvs/dev/local/lib/python2.7/site-packages/pysnmp/proto/rfc3412.py", line 274, in receiveMessage

  File "/home/kalombo/.virtualenvs/dev/local/lib/python2.7/site-packages/pysnmp/smi/builder.py", line 299, in importSymbols

  File "/home/kalombo/.virtualenvs/dev/local/lib/python2.7/site-packages/pysnmp/smi/builder.py", line 270, in loadModules

pysnmp.smi.error.SmiError: MIB file "__SNMPv2-MIB.py[co]" not found in search path

，然后脚本停止。为什么会发生这种情况？

---

解决方案

如果仅当您提高并发级别时才会出现这种情况，则可能的解释是您遇到了平台允许您在任何时候打开的文件数量的限制。每个打开的套接字都计入此限制，打开的"常规"文件(从文件系统)也是如此。

如果您用完了所有允许的文件，则Python无法从磁盘读取模块的源代码，因为该平台不允许它打开它们。

在这种情况下是否会发生这种情况并不是很明显，因为(如果是)pysnMP正在处理真正的异常，并重新引发隐藏详细信息的新异常。

如果这是问题所在，则可以通过提高打开文件限制来解决它。大多数情况下，您可以通过运行以下命令来完成此操作：

$ ulimit -Sn 2048

阅读有关ulimit的更多信息(help ulimit在您的外壳中，位于the internet)，了解有关控制该限制的更多详细信息。