django+tornado實(shí)現(xiàn)實(shí)時(shí)查看遠(yuǎn)程日志的方法
大致思路:
1.利用tornado提供的websocket功能與瀏覽器建立長(zhǎng)連接,讀取實(shí)時(shí)日志并輸出到瀏覽器
2.寫(xiě)一個(gè)實(shí)時(shí)讀取日志的腳本,利用saltstack遠(yuǎn)程執(zhí)行,并把實(shí)時(shí)日志發(fā)往redis中。
3.tornado讀取redis中的信息,發(fā)往瀏覽器。
此過(guò)程用到了redis的發(fā)布和訂閱功能。
先看一下tornado中是如何處理的:
import os
import sys
import tornado.websocket
import tornado.web
import tornado.ioloop
import redis
import salt.client
from tornado import gen
from tornado.escape import to_unicode
from logs.utility import get_last_lines
from logs import settings
class SubWebSocket(tornado.websocket.WebSocketHandler):
"""
此handler處理遠(yuǎn)程日志查看
"""
def open(self, *args, **kwargs):
print("opened")
@gen.coroutine
def on_message(self, message):
# 主機(jī)名,要查看的日志路徑,運(yùn)行腳本的命令這些信息從瀏覽器傳過(guò)來(lái)
hostname, log_path, cmd = message.split("||")
local = salt.client.LocalClient()
r = redis.StrictRedis(host=settings.REDIS_HOST, port=settings.REDIS_PORT,
password=settings.REDIS_PASSWD, db=5)
# 訂閱頻道,服務(wù)器和日志路徑確定一個(gè)頻道
key = settings.LOG_KEY.format(server=hostname.strip(), log_path=log_path.strip())
channel = r.pubsub()
channel.subscribe(key)
# 異步方式執(zhí)行命令,遠(yuǎn)程運(yùn)行腳本
local.cmd_async(hostname, "cmd.run", [cmd])
try:
while True:
data = channel.get_message()
if not data:
# 如果讀取不到消息,間隔一定時(shí)間,避免無(wú)謂的CPU消耗
yield gen.sleep(0.05)
continue
if data["type"] == "message":
line = format_line(data["data"])
self.write_message(line)
except tornado.websocket.WebSocketClosedError:
self.close()
def on_close(self):
global FLAG
FLAG = False
print("closed")
def format_line(line):
line = to_unicode(line)
if "INFO" in line:
color = "#46A3FF"
elif "WARN" in line:
color = "#FFFF37"
elif "ERROR" in line:
color = "red"
elif "CRITICAL" in line:
color = "red"
else:
color = "#FFFFFF"
return "<span style='color:{}'>{}</span>".format(color, line)
class EchoWebSocket(tornado.websocket.WebSocketHandler):
def open(self):
print("WebSocket opened")
@gen.coroutine
def on_message(self, message):
log = message
print "log file: ", log
try:
with open(log, 'r') as f:
for line in get_last_lines(f):
line1 = format_line(line)
self.write_message(line1)
while True:
line = f.readline()
if not line:
yield gen.sleep(0.05)
continue
self.write_message(format_line(line.strip()))
except tornado.websocket.WebSocketClosedError as e:
print e
self.close()
# def check_origin(self, origin):
# print origin, self.request.headers.get("Host")
# # super(EchoWebSocket, self).check_origin()
# return True
def on_close(self):
print("WebSocket closed")
class Application(tornado.web.Application):
def __init__(self):
handlers = [
(r'/log/', MainHandler), # 提供瀏覽頁(yè)面,頁(yè)面中的JS與服務(wù)器建立連接
(r'/log/local', EchoWebSocket), # 處理本地日志實(shí)時(shí)查看,比較簡(jiǎn)單
(r'/log/remote', SubWebSocket), # 處理遠(yuǎn)程日志實(shí)時(shí)查看,稍微復(fù)雜
]
settings = {
"debug": True,
"template_path": os.path.join(os.path.dirname(__file__), "templates"),
"static_path": os.path.join(os.path.dirname(__file__), "static"),
}
super(Application, self).__init__(handlers, **settings)
class MainHandler(tornado.web.RequestHandler):
def get(self):
# 要查看的日志路徑
log = self.get_argument("log", None)
# hostname實(shí)際上是saltstack中這臺(tái)機(jī)器對(duì)應(yīng)的minion id
hostname = self.get_argument("hostname", None)
# 本地日志還是遠(yuǎn)程日志
type = self.get_argument("type", "local")
# 運(yùn)行讀取實(shí)時(shí)日志的腳本,參數(shù)比較多,后面會(huì)有
cmd = self.get_argument("cmd", "")
context = {
"log": log,
"hostname": hostname,
"type": type,
"cmd": cmd,
}
self.render("index.html", **context)
配置文件中主要記錄了redis服務(wù)器的地址等信息
# encoding: utf-8
LOG_KEY = "logs:{server}:{log_path}"
LOG_NAME = "catalina.out"
TAIL_LINE_NUM = 20
REDIS_HOST = "127.0.0.1"
REDIS_PORT = "6379"
REDIS_PASSWD = None
REDIS_EXPIRE = 300
try:
from local_settings import *
except ImportError:
pass
index.html的內(nèi)容如下:
<html>
<head>
<link href="{{ static_url('public/css/public.css') }}" rel="external nofollow" rel="stylesheet" />
<link href="{{ static_url('kylin/css/style.css') }}" rel="external nofollow" rel="stylesheet" />
</head>
<body style="background:#000000">
<div style="margin-left:10px;">
<pre id="id-content">
</pre>
<div id="id-bottom"></div>
<input type="hidden" id="id-log" value="{{ log }}" />
<input type="hidden" id="id-type" value="{{ type }}" />
<input type="hidden" id="id-hostname" value="{{ hostname }}" />
<input type="hidden" id="id-cmd" value="{{ cmd }}" />
<div class="btns btns_big">
<button type="button" class="query_btn cancle" id="id-stop">Stop</button>
<button type="button" class="query_btn commit" id="id-start">Start</button>
</div>
</div>
<script type="text/javascript" src="{{ static_url('js/jquery-1.11.3.min.js') }}"></script>
<script type="text/javascript">
var log_name = $("#id-log").val();
var type = $("#id-type").val();
var hostname = $("#id-hostname").val();
var cmd = $("#id-cmd").val();
// 初始化websocket對(duì)象
var ws = new WebSocket("ws://{{ request.host }}/log/" + type);
ws.onopen = function(){
if (type === "local"){
ws.send(log_name);
} else {
// 建立連接后把相關(guān)信息發(fā)往服務(wù)器,對(duì)應(yīng)上面的SubWebSocket
ws.send(hostname + "||" + log_name + "||" + cmd);
}
};
var get_message = function(evt){
$("#id-content").append(evt.data + "\n");
document.getElementById("id-bottom").scrollIntoView()
};
ws.onmessage = get_message;
// 兩個(gè)按鈕控制日志的輸出,如果看到需要的日志信息,可以暫停日志的輸出,
// 之后可以繼續(xù)啟動(dòng)日志的輸出
$("#id-stop").click(function(){
ws.onmessage = function(){};
})
$("#id-start").click(function(){
ws.onmessage = get_message;
})
</script>
</body>
</html>
這個(gè)tornado僅僅是提供了實(shí)時(shí)日志的服務(wù),實(shí)際項(xiàng)目使用的是django,django中要做的其實(shí)很簡(jiǎn)單,提供log_name,hostname,type,cmd等四個(gè)參數(shù)。
下面看一個(gè)實(shí)例:
class LogView(KylinView):
# 實(shí)時(shí)讀取日志的腳本,事先使用saltstack批量傳到各臺(tái)服務(wù)器上
client_path = "/tmp/logtail.py"
def get(self, request):
minion_id = request.GET.get("minion_id")
context = {
"minion_id": minion_id,
"tail_log_url": settings.TAIL_LOG_URL,
}
return render(request, "cmdb/log_view.html", context)
def post(self, request):
minion_id = request.POST.get("minion_id")
log_path = request.POST.get("log_path")
if not log_path:
return JsonResponse({"success": False, "message": "請(qǐng)?zhí)顚?xiě)日志路徑"})
try:
# 制定一開(kāi)始讀取的行數(shù)
line_count = request.POST.get("line_count")
except (TypeError, ValueError):
return JsonResponse({"success": False, "message": "請(qǐng)輸入正確的行數(shù)"})
local = salt.client.LocalClient()
# 確保saltstack能連通并且日志文件存在
ret = local.cmd(minion_id, "file.file_exists", [log_path])
if minion_id not in ret:
return JsonResponse({"success": False, "message": "服務(wù)器無(wú)法連通"})
if not ret[minion_id]:
return JsonResponse({"success": False, "message": "日志文件不存在"})
# 組成命令的各個(gè)參數(shù),redis信息需要和tornado配置文件中的redis信息一致
cmd = "{} {} {} {} {} {} {} {}".format(
settings.PYTHON_BIN, self.client_path, minion_id, log_path, line_count, settings.REDIS_HOST,
settings.REDIS_PORT, settings.REDIS_PASSWD)
# settings.TAIL_LOG_URL是tornado中MainHandler對(duì)應(yīng)的url,把其它幾個(gè)
# 參數(shù)組合成最終的URL,直接訪問(wèn)這個(gè)URL就可以在瀏覽器中實(shí)時(shí)讀取日志了。
url = "{}?type=remote&log={}&hostname={}&cmd={}".format(
settings.TAIL_LOG_URL, log_path, minion_id, cmd)
# 這一步的操作確保同一個(gè)日志文件只有一個(gè)腳本在讀取,避免日志信息重復(fù),這一步
# 也很重要,必不可少
local.cmd(minion_id, "cmd.run",
["kill `ps aux|grep logtail.py|grep %s|grep -v grep|awk '{print $2}'`" % (log_path,)])
return JsonResponse({"success": True, "url": url})
下面來(lái)看看logtail.py的實(shí)現(xiàn):
# encoding: utf-8
from __future__ import unicode_literals, division
import math
import time
import sys
import socket
import signal
import redis
FLAG = True
def get_last_lines(f, num=10):
"""讀取文件的最后幾行
"""
size = 1000
try:
f.seek(-size, 2)
except IOError: # 文件內(nèi)容不足size
f.seek(0)
return f.readlines()[-num:]
data = f.read()
lines = data.splitlines()
n = len(lines)
while n < num:
size *= int(math.ceil(num / n))
try:
f.seek(-size, 2)
except IOError:
f.seek(0)
return f.readlines()[-num:]
data = f.read()
lines = data.splitlines()
n = len(lines)
return lines[-num:]
def process_line(r, channel, line):
r.publish(channel, line.strip())
def sig_handler(signum, frame):
global FLAG
FLAG = False
# 收到退出信號(hào)后,以比較優(yōu)雅的方式終止腳本
signal.signal(signal.SIGTERM, sig_handler)
# 為了避免日志輸出過(guò)多,瀏覽器承受不住,設(shè)置5分鐘后腳本自動(dòng)停止
signal.signal(signal.SIGALRM, sig_handler)
signal.alarm(300)
def get_hostname():
return socket.gethostname()
def force_str(s):
if isinstance(s, unicode):
s = s.encode("utf-8")
return s
def tail():
password = sys.argv[6]
if password == "None":
password = None
r = redis.StrictRedis(host=sys.argv[4], port=sys.argv[5], password=password, db=5)
log_path = sys.argv[2]
line_count = int(sys.argv[3])
# 往redis頻道發(fā)送實(shí)時(shí)日志
channel = "logs:{hostname}:{log_path}".format(hostname=sys.argv[1], log_path=log_path)
with open(log_path, 'r') as f:
last_lines = get_last_lines(f, line_count)
for line in last_lines:
process_line(r, channel, force_str(line))
try:
while FLAG: # 通過(guò)信號(hào)控制這個(gè)變量,實(shí)現(xiàn)優(yōu)雅退出循環(huán)
line = f.readline()
if not line:
time.sleep(0.05)
continue
process_line(r, channel, line)
except KeyboardInterrupt:
pass
print("Exiting...")
if __name__ == "__main__":
if len(sys.argv) < 6:
print "Usage: %s minion_id log_path host port redis_pass"
exit(1)
tail()
到此為止,整個(gè)實(shí)時(shí)讀取遠(yuǎn)程日志的流程就講完了。
github: https://github.com/tuxinhang1989/logs
以上這篇django+tornado實(shí)現(xiàn)實(shí)時(shí)查看遠(yuǎn)程日志的方法就是小編分享給大家的全部?jī)?nèi)容了,希望能給大家一個(gè)參考,也希望大家多多支持腳本之家。
相關(guān)文章
python調(diào)用接口的4種方式代碼實(shí)例
這篇文章主要介紹了python調(diào)用接口的4種方式代碼實(shí)例,文中通過(guò)示例代碼介紹的非常詳細(xì),對(duì)大家的學(xué)習(xí)或者工作具有一定的參考學(xué)習(xí)價(jià)值,需要的朋友可以參考下2019-11-11
在Python 2.7即將停止支持時(shí),我們?yōu)槟銕?lái)了一份python 3.x遷移指南
這篇文章主要介紹了在Python 2.7即將停止支持時(shí)我們?yōu)槟銣?zhǔn)備了一份python 3.x遷移指南的相關(guān)資料,需要的朋友可以參考下2018-01-01
Python函數(shù)中的函數(shù)(閉包)用法實(shí)例
這篇文章主要介紹了Python函數(shù)中的函數(shù)(閉包)用法,結(jié)合實(shí)例形式分析了Python閉包的定義與使用技巧,需要的朋友可以參考下2016-03-03
Python利用itchat庫(kù)向好友或者公眾號(hào)發(fā)消息的實(shí)例
今天小編就為大家分享一篇Python利用itchat庫(kù)向好友或者公眾號(hào)發(fā)消息的實(shí)例,具有很好的參考價(jià)值,希望對(duì)大家有所幫助。一起跟隨小編過(guò)來(lái)看看吧2019-02-02
Django cookie和session的應(yīng)用場(chǎng)景及如何使用
今天我們來(lái)重點(diǎn)看下Django中session和cookie的用法吧。我們會(huì)介紹cookie和session的工作原理,還會(huì)分享實(shí)際應(yīng)用的案例。2021-04-04
Python7個(gè)爬蟲(chóng)小案例詳解(附源碼)下篇
這篇文章主要介紹了Python7個(gè)爬蟲(chóng)小案例詳解(附源碼)上篇,本文章內(nèi)容詳細(xì),通過(guò)案例可以更好的理解爬蟲(chóng)的相關(guān)知識(shí),七個(gè)例子分為了三部分,本次為下篇,共有三道題,需要的朋友可以參考下2023-01-01
13個(gè)Pandas實(shí)用技巧,助你提高開(kāi)發(fā)效率
這篇文章主要介紹了13個(gè)Pandas實(shí)用技巧,幫助你提高python開(kāi)發(fā)的效率,感興趣的朋友可以了解下2020-08-08
python用tkinter開(kāi)發(fā)的掃雷游戲
非常簡(jiǎn)單的實(shí)現(xiàn),所以并沒(méi)有那么多繁瑣得步驟,對(duì)于這種簡(jiǎn)單得h5游戲來(lái)說(shuō),用python可以很容易就設(shè)計(jì)出來(lái)。下面就來(lái)看看實(shí)現(xiàn)步驟2021-06-06

