-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy path02-05-15-#sdnds-tw.txt
193 lines (193 loc) · 21.1 KB
/
02-05-15-#sdnds-tw.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
13:49 +0800 <sdnds-slackbot>: <hellblazzar>: 不好意思上來問個問題,最近做實驗遇到了奇怪的狀況,在Controller端可以看見host,但是實際上各host間及controller與host卻無法ping通,也嘗試過直接下flow到switch上,依然是不行,請問這種狀況該怎麼解決呢?(我的環境是OpenWrt+OpenvSwitch配上Floodlight)
13:56 +0800 <sdnds-slackbot>: <john-lin>: Controller 本來就ping不到host啊
13:57 +0800 <sdnds-slackbot>: <hellblazzar>: 那host間跟host ping controller也是不允許嗎?
14:00 +0800 <sdnds-slackbot>: <hellblazzar>: 還有就是如果controller不能ping host,這樣controller是怎麼看到host的呢?
14:00 +0800 <sdnds-slackbot>: <john-lin>: Host間可以互通 但要在flow entry下的正確的前提之下
14:01 +0800 <sdnds-slackbot>: <john-lin>: Controller 只管switch上的每一個link status
14:02 +0800 <sdnds-slackbot>: <hellblazzar>: 所以在host互ping前,我必須先下好flow entry囉?
14:02 +0800 <sdnds-slackbot>: <john-lin>: 對啊
14:03 +0800 <sdnds-slackbot>: <john-lin>: 沒有entry 封包怎麼forward
14:07 +0800 <sdnds-slackbot>: <hellblazzar>: 但是在switch不知道怎麼導封包的狀況下,照理來說會塞給floodlight controller,再藉由floodlight controller來導吧?
14:08 +0800 <sdnds-slackbot>: <john-lin>: 會
14:08 +0800 <sdnds-slackbot>: <john-lin>: 所以你要在controller 處理這事
14:09 +0800 <sdnds-slackbot>: <john-lin>: 但是你還是要有一筆entry是forward到controller
14:10 +0800 <sdnds-slackbot>: <john-lin>: Table miss
14:11 +0800 <sdnds-slackbot>: <hellblazzar>: 可是openvswitch不是本來就知道怎麼到controller嗎?還是它只是知道controller是誰,但是不知道怎麼到controller
14:12 +0800 <sdnds-slackbot>: <john-lin>: 1.3 預設是把table miss drop
14:13 +0800 <sdnds-slackbot>: <john-lin>: 除非你把table miss 改為forward to controller
14:16 +0800 <sdnds-slackbot>: <hellblazzar>: 歐歐,那這部分是我還不夠熟openflow的protocol,我再把openflow多看幾次好了@@
14:16 +0800 <sdnds-slackbot>: <hellblazzar>: 謝謝john-lin大
14:16 +0800 <sdnds-slackbot>: <john-lin>: 不會
15:11 +0800 <sdnds-slackbot>: <hellblazzar>: 不好意思john-lin大,剛剛忘了說我是用OpenFlow1.0,而就我查證過關於openvswitch關於table miss的處理,官方上是說針對openflow1.0的部分會是在table miss後導向controller,而floodlight本身預設會去做forward封包的動作,但是這樣就我就不了解為什麼互ping會失敗這件事了
15:12 +0800 <sdnds-slackbot>: <john-lin>: 先確認一下你的 table miss 確實有寫進去
15:14 +0800 <sdnds-slackbot>: <hellblazzar>: 是指從controller端去把封包撈出來看嗎?
15:15 +0800 <sdnds-slackbot>: <hellblazzar>: 我試試
15:16 +0800 <sdnds-slackbot>: <john-lin>: 直接在switch上用OVS的dump flows指令
15:16 +0800 <sdnds-slackbot>: <hellblazzar>: 歐歐
15:20 +0800 <sdnds-slackbot>: <hellblazzar>: 我這邊看到是空的,這樣是代表沒有table miss的動作嗎?
15:20 +0800 <sdnds-slackbot>: <john-lin>: 是的
15:21 +0800 <sdnds-slackbot>: <hellblazzar>: 不過 確認table miss不是使用 dump-table-features這個指令嗎??
15:21 +0800 <sdnds-slackbot>: <john-lin>: 沒有features吧
15:21 +0800 <sdnds-slackbot>: <hellblazzar>: 是說 ovs會吐回來限定1.3之後才省用這個指令的錯誤訊息
15:21 +0800 <sdnds-slackbot>: <hellblazzar>: @@
15:22 +0800 <sdnds-slackbot>: <john-lin>: ovs-ofctl dump-flows
15:22 +0800 <sdnds-slackbot>: <john-lin>: 指定protocol -O openflow10
15:26 +0800 <sdnds-slackbot>: <hellblazzar>: 現在長這樣
15:26 +0800 <sdnds-slackbot>: <hellblazzar>: 不知道為什麼又多2到flow entry了@@
15:28 +0800 <sdnds-slackbot>: <john-lin>: 先指定一下protocol吧
15:28 +0800 <sdnds-slackbot>: <john-lin>: -O
15:28 +0800 <sdnds-slackbot>: <hellblazzar>: 是說,這樣看來我的openvswitch把table miss的本機當成192.168.1.1了
15:28 +0800 <sdnds-slackbot>: <hellblazzar>: 好的
15:30 +0800 <sdnds-slackbot>: <hellblazzar>: 請問一下這兩條就是table miss了嗎?
15:30 +0800 <sdnds-slackbot>: <john-lin>: 不是
15:30 +0800 <sdnds-slackbot>: <hellblazzar>: 歐歐
15:31 +0800 <sdnds-slackbot>: <john-lin>: Table miss 是match 所有封包 priority 最低的那條
15:32 +0800 <sdnds-slackbot>: <john-lin>: Action 是controller
15:34 +0800 <sdnds-slackbot>: <hellblazzar>: 那麼,按1.0的規定來看,ovs會去生成table miss entry嗎?還是基本上我要自己來下?
15:35 +0800 <sdnds-slackbot>: <john-lin>: 是floodlight 自己要去handle
15:36 +0800 <sdnds-slackbot>: <john-lin>: 但是我不知道floodlight是不是有在framework 做這件事 還是要開發者自己去寫table miss
15:37 +0800 <sdnds-slackbot>: <john-lin>: 這可能要問 floodlight 的核心開發者 @hwchiu
15:37 +0800 <sdnds-slackbot>: <hellblazzar>: 恩...在我把ovs接上controller後,每次都會自動生成這兩條flow倒是有的
15:38 +0800 <sdnds-slackbot>: <hellblazzar>: 只是我的ovs-bridge ip卻是192.168.5.1
15:38 +0800 <sdnds-slackbot>: <hellblazzar>: 明顯不合現狀呢...
15:39 +0800 <sdnds-slackbot>: <john-lin>: 我也不知道你那兩條是怎麼生出來的 但你確實需要一條table miss 不然沒辦法繼續做下去
15:40 +0800 <sdnds-slackbot>: <john-lin>: 有沒有那兩條其實無所謂 有table miss的entry 你就能在controller 去處理packet in 的事件
15:42 +0800 <sdnds-slackbot>: <hellblazzar>: 原來如此...那麼我自己手動建一條好了
15:43 +0800 <sdnds-slackbot>: <hellblazzar>: 現在看來也只有這個用做法來因應一下現狀@@
15:43 +0800 <sdnds-slackbot>: <rascov>: 1.0 應該不用 controller 手動下
15:43 +0800 <sdnds-slackbot>: <rascov>: 應該有一條 hidden entry
15:43 +0800 <sdnds-slackbot>: <rascov>: 若我沒記錯的話
15:44 +0800 <sdnds-slackbot>: <hellblazzar>: 歐歐
15:45 +0800 <sdnds-slackbot>: <hellblazzar>: rascov大,那我要怎麼知道hidden entry有沒有確實被建立起來了
15:45 +0800 <sdnds-slackbot>: <rascov>: 你們剛剛檢查過hidden entry了嗎
15:46 +0800 <sdnds-slackbot>: <hellblazzar>: 還沒有,這部分也能透過ovs查到嗎?
15:46 +0800 <sdnds-slackbot>: <hellblazzar>: 我試試
15:46 +0800 <sdnds-slackbot>: <rascov>: 我現在在外面無法告訴你
15:46 +0800 <sdnds-slackbot>: <rascov>: 你Google ovs hidden entry 通告可以找到小飛機部落格某一篇
15:46 +0800 <sdnds-slackbot>: <rascov>: http://Roan.logdown.com 好像在這
15:46 +0800 <sdnds-tw_Bot>: ==> roan's Blog
15:47 +0800 <sdnds-slackbot>: <hellblazzar>: 好的,我試試看
15:47 +0800 <sdnds-slackbot>: <john-lin>: 可以
15:48 +0800 <sdnds-slackbot>: <john-lin>: ovs-appctl bridge/dump-flows ovs-br
15:48 +0800 <sdnds-slackbot>: <hellblazzar>: 歐歐 要用到appctl呀
15:49 +0800 <sdnds-slackbot>: <john-lin>: 但是我在ryu上 table miss的flow 不是hidden 的
15:49 +0800 <sdnds-slackbot>: <john-lin>: 你還是先check吧
15:50 +0800 <sdnds-slackbot>: <rascov>: 我OVS開起來沒有任何entry就預設會回controller (1.0)
15:50 +0800 <sdnds-slackbot>: <hellblazzar>: 歐歐 我在FAQ裡有看到了 在controller的章節
15:54 +0800 <sdnds-slackbot>: <hellblazzar>: 照結果看來,就和rascov大說的那樣,有hidden flow存在
15:54 +0800 <sdnds-slackbot>: <hellblazzar>: 不過行為卻是normal而非controller
15:55 +0800 <sdnds-slackbot>: <hellblazzar>: @@
15:56 +0800 <sdnds-slackbot>: <john-lin>: 有forward to controller 倒數第三筆
15:59 +0800 <sdnds-slackbot>: <hellblazzar>: 歐歐0.0 我眼殘了
16:00 +0800 <sdnds-slackbot>: <hellblazzar>: 但是dump-tables只有到253
16:01 +0800 <sdnds-slackbot>: <hellblazzar>: 所以這254指的是全部miss才會到達的一張table囉?
16:02 +0800 <sdnds-slackbot>: <john-lin>: 嗯 他預設 pipeline 應該會跑到最後一張
16:03 +0800 <sdnds-slackbot>: <hellblazzar>: 只是...就結果來看,我還是不懂為什麼ping不通這件事,如果應有的規則都有了,照理說網路會正常運作才是
16:04 +0800 <sdnds-slackbot>: <hellblazzar>: 還是問題其實是在floodlight?
16:04 +0800 <sdnds-slackbot>: <john-lin>: 你現在只確認他會forward 到controller 還沒有確認controller接到packet in 該做什麼事
16:05 +0800 <sdnds-slackbot>: <hellblazzar>: 恩
16:05 +0800 <sdnds-slackbot>: <hellblazzar>: 那我重新檢查一下floodlight這邊的狀況
16:06 +0800 <sdnds-slackbot>: <john-lin>: 應該看一下floodlight 的application 有沒有做learning bridge
16:06 +0800 <sdnds-slackbot>: <john-lin>: 不然他還是ping不通
16:17 +0800 <sdnds-slackbot>: <hellblazzar>: 不過...就我看到的來說,在floodlight核心的元件中,確實有著forwarding的部分,就在Forwarding.java內有規範說如果有paket in的封包近來,會先檢驗目標是不是在同一台ovs上,在決定要不要flood,但是我啟動時是直接載入預設的模組,Forwarding這個module是會被載進來才對
16:18 +0800 <sdnds-slackbot>: <hellblazzar>: 而forward這部分就不是以application的方式存在了
16:20 +0800 <sdnds-slackbot>: <hellblazzar>: 只是這樣子的話,問題感覺不像在ovs到controller這部分,比較像在我的ovs上?
16:22 +0800 <sdnds-slackbot>: <hellblazzar>: 感覺比較接近防火牆之類的,我在想想好了@@
16:22 +0800 <sdnds-slackbot>: <john-lin>: 那意思就是已經預載了 modules 你要不要嘗試先把packet in的封包資訊打出來看
16:24 +0800 <sdnds-slackbot>: <hellblazzar>: 那我實際把封包撈出來看好了,確認一下封包有沒有真的進來controller,還是在這中間就被吃掉了
17:40 +0800 <sdnds-slackbot>: <hellblazzar>: john-lin大,我現在測出來的狀況是這樣:在controller上我用wireshark監聽controller的網卡,當有host ping controller時,我有看到controller有收到來自ovs的詢問的訊息,而且controller也有廣播自己的mac回去,但是ping controller的host並沒有收到(這台host上的wireshark只有看到他自己詢問時的廣播封包),這證明controller的運作是正常的,而且%E
17:43 +0800 <sdnds-slackbot>: <hellblazzar>: 而在controller上ettercap看不到ping controller的host,nmap也是失敗
17:43 +0800 <sdnds-slackbot>: <john-lin>: 你為什麼想用host ping controller?
17:44 +0800 <sdnds-slackbot>: <hellblazzar>: 因為我想從別台host下rest api給controller
17:44 +0800 <sdnds-slackbot>: <john-lin>: Data plane 跟control plane 是分離的
17:45 +0800 <sdnds-slackbot>: <john-lin>: Dataplane沒辦法去access controller
17:45 +0800 <sdnds-slackbot>: <john-lin>: 你應該要在control plane 那邊去發rest
17:47 +0800 <sdnds-slackbot>: <hellblazzar>: 那這樣子 ,一般的利用north bound api的application 或service和controller間連接時,不能透過openflow嗎?
17:47 +0800 <sdnds-slackbot>: <hellblazzar>: 還是這時是把controller當成普通的host來看?
17:48 +0800 <sdnds-slackbot>: <hellblazzar>: 因為我最終是想利用SDN的特性進行VoIP
17:49 +0800 <sdnds-slackbot>: <hellblazzar>: 所以勢必會利用到call controller的rest api這部分
17:49 +0800 <sdnds-slackbot>: <john-lin>: 用openflow protocol 是在southbound
17:51 +0800 <sdnds-slackbot>: <hellblazzar>: 那麼,我openflow網路內的host要怎麼存取到northbound的services呢?
17:52 +0800 <sdnds-slackbot>: <john-lin>: 照SDN 架構來說Data plane 應該是存取不到controller的
17:54 +0800 <sdnds-slackbot>: <john-lin>: 你應該是在controller那邊有一隻程式去發rest api跟controller溝通
17:55 +0800 <sdnds-slackbot>: <john-lin>: 如果硬要在dataplane 去access controller 也不是不行 你變成要有一台第三方的 host同時掛在data plane 跟control plane上
19:04 +0800 <sdnds-slackbot>: <hellblazzar>: 所以northbound application基本上是以這兩種形式存在囉?
19:05 +0800 <sdnds-slackbot>: <hellblazzar>: 第二種感覺比較清楚,但john-line大說的第一種(右邊那張)感覺比較模糊,不知道是不是那樣?
19:06 +0800 <sdnds-slackbot>: <john-lin>: 看不出來你綠色那個是在做什麼
19:06 +0800 <sdnds-slackbot>: <john-lin>: 還有service 不一定要access dataplane啊
19:08 +0800 <sdnds-slackbot>: <rascov>: 覺得這設計有點奇怪
19:08 +0800 <sdnds-slackbot>: <rascov>: 應該是網路要偵測host的行為做應對,不應該host直接跟controller溝通
19:09 +0800 <sdnds-slackbot>: <rascov>: 偵測到host送出 VoIP 流量controller再去設定網路就好了啊
19:10 +0800 <sdnds-slackbot>: <john-lin>: 嗯 想知道為什麼一定要讓host 跟controller 接
19:11 +0800 <sdnds-slackbot>: <rascov>: 對啊 可以描述一下你想達到什麼目標嗎
19:20 +0800 <sdnds-slackbot>: <hellblazzar>: 我想達到的目標是這樣:我的VoIP server(我是使用 SIP協定)也建在OpenFlow網路中,當我底下host打電話時,就必須向server索取受話的host的IP,這時SIP Server便利用rest-api要求controller建立一條連通兩台Host的通道(以下flow的方式達成),接下來利用SIP p2p的特性,以後通話就不需要再經過server,而是在底下的OpenFlow network��%8
19:20 +0800 <sdnds-slackbot>: <hellblazzar>: 還是我的架構有問題呢?因為照我想這是最直觀的方式了
19:23 +0800 <sdnds-slackbot>: <rascov>: 你有沒有辦法聽 SIP server 傳回給 client 的封包就決定路徑
19:23 +0800 <sdnds-slackbot>: <rascov>: 把所有SIP control packet In 給controller
19:26 +0800 <sdnds-slackbot>: <rascov>: 若你的流量有加密,或是這個解法不行
19:26 +0800 <sdnds-slackbot>: <rascov>: Host也有辦法跟controller溝通啦,可參考ONOS BGP server 跟controller app溝通的方式
19:28 +0800 <sdnds-slackbot>: <hellblazzar>: 所以是,host與host間p2p後,controller監聽到這兩者間有VoIP封包,就自動去建Flow了?
19:33 +0800 <sdnds-slackbot>: <john-lin>: 你在controller 可以接到所有SIP 的封包透過packet in 然後decode SIP 就能做到不過加密就很難了。如果SIP server 放在control plane呢?兩個interface
19:34 +0800 <sdnds-slackbot>: <john-lin>: 等於你畫的圖service 就是SIP server
19:37 +0800 <sdnds-slackbot>: <hellblazzar>: 可是如果把Server放在Control Plane,那麼在接通前的"向Server要IP"這一步基本上不就會出問題了嗎?
19:39 +0800 <sdnds-slackbot>: <john-lin>: Hosts 向server要IP 透過server在下面的interface
19:39 +0800 <sdnds-slackbot>: <john-lin>: 我剛舉的例子是SIP server要有兩個interface
19:45 +0800 <sdnds-slackbot>: <hellblazzar>: 恩...好抽象@@,是指在controller內加一個module用來解析SIP封包嗎?
19:47 +0800 <sdnds-slackbot>: <john-lin>: 剛剛 @rascov 舉的例子是 會讓controller預先下一個flow entry是forward 所有的sip封包到controller 然後你要在controller那邊 decode就能做到 這個只要把controller放在data plane就可以完成
19:48 +0800 <sdnds-slackbot>: <john-lin>: 我剛剛舉的例子是 按照你原先的圖 把你的service方塊換成SIP server 你剛畫了兩條線,所以實體上會有兩個 network interfaces
19:49 +0800 <sdnds-slackbot>: <john-lin>: 就是你剛放上來的 untitled 3.png 圖
19:57 +0800 <sdnds-slackbot>: <hellblazzar>: 恩恩,我了解了,因為我最後一張圖也是照著這個想法來畫的
20:02 +0800 <sdnds-slackbot>: <hwchiu>: 有重點嬤XD
20:03 +0800 <sdnds-slackbot>: <john-lin>: Floodlight 專家來了XD
20:03 +0800 <sdnds-slackbot>: <hwchiu>: hellblazzar: 你用什麼版本的floodlight?
20:04 +0800 <sdnds-slackbot>: <hwchiu>: 0.9? 1.0? 1.1?
20:05 +0800 <sdnds-slackbot>: <hellblazzar>: 恩...不過不好意思john-lin大,我剛剛又針對host與host間互ping這件事做一次測試(這是目標是OpenFlow網路中的host,不再是controller),結果還是不行,我的發出端依然收不到目標端傳回的arp@@
20:05 +0800 <sdnds-slackbot>: <hellblazzar>: 歐歐
20:05 +0800 <sdnds-slackbot>: <hellblazzar>: 是hwchui大
20:05 +0800 <sdnds-slackbot>: <hellblazzar>: 你好^^ 我用的是1.0版的
20:06 +0800 <sdnds-slackbot>: <hwchiu>: 就我所知
20:06 +0800 <sdnds-slackbot>: <hwchiu>: 1.3後, default的行為改變
20:06 +0800 <sdnds-slackbot>: <hwchiu>: 所以floodlight後來在switch連線後會送一個flow entry: output=controller當作table miss entry
20:06 +0800 <sdnds-slackbot>: <hwchiu>: 我不確定你的版本有沒有那段code
20:06 +0800 <sdnds-slackbot>: <hwchiu>: 你要不要試試看目前的1.1
20:07 +0800 <sdnds-slackbot>: <rascov>: 他跑OF 1.0
20:07 +0800 <sdnds-slackbot>: <hwchiu>: 哦哦
20:07 +0800 <sdnds-slackbot>: <hellblazzar>: 我是使用OF1.0
20:07 +0800 <sdnds-slackbot>: <hellblazzar>: 恩恩
20:07 +0800 <sdnds-slackbot>: <hwchiu>: 有辦法開個環境讓我去看看嗎XD
20:08 +0800 <sdnds-slackbot>: <john-lin>: @hellblazzar: 你在controller上的問題要不要再重複一下
20:08 +0800 <sdnds-slackbot>: <hellblazzar>: 好的
20:22 +0800 <sdnds-slackbot>: <hellblazzar>: 基本上就是我想è¦�在我建立的環境下(Floodlight 1.0 + OpenvSwitch on OpenWrt,採OpenFlow1.0),讓兩å�°Host互相ping通,而就我原先的想法,在沒有flow entryçš„ç‹€æ³�下,ovs會將å°�包導å�‘controller,並藉由controller來引導,但是å�»å¤±æ•—了,所以我想è¦�é‡�清å•�題的癥çµ�點,一是關於floodlight controller,也就是floodlight是å�¦æœƒæœ‰å¼•å°Žå°�包的行為,二就是最差的ç
20:23 +0800 <sdnds-slackbot>: <hellblazzar>: 雖然就今天與john-lin大討論與實驗的結果看來是二,但還是想聽聽hwchiu大您的意見
20:24 +0800 <sdnds-slackbot>: <hwchiu>: 先確認幾個點 1) version of OVS, 2)能否包一下 floodlight整個專案借我測試看看
20:27 +0800 <sdnds-slackbot>: <hellblazzar>: 歐歐,我是使用openwrt官方釋出的ovs 2.3.9的packet,floodlight我是使用乾淨、沒修改過的1.0版,請等我一下
20:32 +0800 <sdnds-slackbot>: <hwchiu>: 哦哦
20:34 +0800 <sdnds-slackbot>: <hwchiu>: 你可以先在ovs手動下flow把兩個host的封包互導試試看? 看看沒有controller的介入下,你的ovs & host network 的網路環境是否正確?
20:36 +0800 <sdnds-slackbot>: <hellblazzar>: <https://drive.google.com/file/d/0BxzGJmUno9p_TjBCc3RTd0R4emM/view?usp=sharing>
20:37 +0800 <sdnds-slackbot>: <hellblazzar>: 歐歐,這步我倒沒試過,我做看看
20:56 +0800 <sdnds-slackbot>: <hellblazzar>: 不行,還是失敗了...我fail mode使用standalone,flow加了下面這兩條
20:57 +0800 <sdnds-slackbot>: <hwchiu>: show一下
20:58 +0800 <sdnds-slackbot>: <hwchiu>: ovs-dpctl dump-flows
20:58 +0800 <sdnds-slackbot>: <hwchiu>: 的結果
20:58 +0800 <sdnds-slackbot>: <hwchiu>: 感覺是2) 你的ovs + openwrt 就有問題
21:00 +0800 <sdnds-slackbot>: <hwchiu>: 你ofctl dump-flows可以看到entry都沒有被match過
21:00 +0800 <sdnds-slackbot>: <hwchiu>: 所以封包根本沒有走到ovs上
21:04 +0800 <sdnds-slackbot>: <rascov>: 牛人突破了盲點
21:05 +0800 <sdnds-slackbot>: <john-lin>: 這跟下午看的entry不一樣啊~
21:19 +0800 <sdnds-slackbot>: <hellblazzar>: 恩...這是我開始丟封包前的狀態,在我開始丟封包後,接收的host會看到詢問接收者的arp,然後接收端也有回應自己mac位址,但是回應的訊息卻傳不回發送端的host,當hwchiu大要我看ovs-dpctl dump-flow前,我就被實驗室管理員趕回家惹QAQ
21:20 +0800 <sdnds-slackbot>: <hellblazzar>: 是說,下午的entry在我把controller拔掉之後就不見了
21:21 +0800 <sdnds-slackbot>: <hellblazzar>: 不過我依然贊同hwchiu大的說法
21:21 +0800 <sdnds-slackbot>: <john-lin>: controller 斷線的話 flow 不久後就會清掉
21:21 +0800 <sdnds-slackbot>: <hellblazzar>: 恩恩
21:22 +0800 <sdnds-slackbot>: <hellblazzar>: 只是,現在就是openwrt設定上的問題了,感覺這會比較棘手些@@
21:25 +0800 <sdnds-slackbot>: <pichuang>: 你先不要把你的網卡 add-port
21:25 +0800 <sdnds-slackbot>: <pichuang>: 先測一般的狀況
21:26 +0800 <sdnds-slackbot>: <pichuang>: <http://roan.logdown.com/posts/238771-openvswitch-debug-enviroment>
21:26 +0800 <sdnds-tw_Bot>: ==> OpenvSwitch Debug$ Enviroment « roan's Blog
21:26 +0800 <sdnds-slackbot>: <pichuang>: 這篇從上面照著慢慢確認下來
21:28 +0800 <sdnds-slackbot>: <pichuang>: openwrt 如果不裝 ovs 的話 可以ping到外面去嗎
21:44 +0800 <sdnds-slackbot>: <hellblazzar>: 好的,我明天再去實驗室試試看,只是我建構的環境和pichuang大你不一樣的是,我沒有利用額外的switch來配給ovs bridge一個ip`,
21:44 +0800 <sdnds-slackbot>: <hellblazzar>: 而是直接建構一個純內網的環境
21:47 +0800 <sdnds-slackbot>: <hellblazzar>: 只是不知道這樣會不會有影響,主要是差在gateway的部分,因為沒有gateway,所以我直接把gateway設為ovs bridge網域下的 .254,但實際上有沒有設都是一樣的
21:47 +0800 <sdnds-slackbot>: <hellblazzar>: 如果有差別基本上是在這部分
21:48 +0800 <sdnds-slackbot>: <hellblazzar>: 因為其實我在架設時也有參照到pichuang大的作法