Issue 324560043: Admission Control
|
Created:
6 years, 7 months ago by Vasileios Papageorgiou Modified:
6 years, 4 months ago Reviewers:
Michail.Xirouchakis, dkostic, younes.tahri3, paul.busi38, michail.xirouchakis, alireza.farshin CC:
2016shana2333, engr.sikandar.iqbal, sdnik2200ht17_ict.kth.se; Visibility:
Public. |
Descriptionminor change
BUG=Rest Api on the nfm
Patch Set 1 #Patch Set 2 : Nfm posts json data to DM #
Total comments: 2
Patch Set 3 : minor change #
Total comments: 1
Patch Set 4 : Get_Request #Patch Set 5 : Fix Get with arguments #
Total comments: 2
Patch Set 6 : Admission Control logic #Patch Set 7 : Admission Control #Patch Set 8 : Deny function of admission Control #Patch Set 9 : Update on Admission logic #
Total comments: 1
Patch Set 10 : Hard time out for adding flow #MessagesTotal messages: 14
Check feedback regarding the delta calculation between two consecutive NFM measurements. It needs to be per switch. https://codereview.appspot.com/324560043/diff/40001/nfm/SimpleNFM13.py File nfm/SimpleNFM13.py (right): https://codereview.appspot.com/324560043/diff/40001/nfm/SimpleNFM13.py#newcode40 nfm/SimpleNFM13.py:40: self.delta_ingress = 0 The delta must be calculated per switch (datapath id). Here it seems like you are using the same delta for all switches which will obviously give you the wrong measurement.
Sign in to reply to this message.
Code review https://codereview.appspot.com/324560043/diff/20001/nfm/SimpleNFM13.py File nfm/SimpleNFM13.py (right): https://codereview.appspot.com/324560043/diff/20001/nfm/SimpleNFM13.py#newcode2 nfm/SimpleNFM13.py:2: delete useless space between different import https://codereview.appspot.com/324560043/diff/20001/nfm/SimpleNFM13.py#newcod... nfm/SimpleNFM13.py:140: Update code with new version
Sign in to reply to this message.
Sign in to reply to this message.
Hello Vasileios, I just added a suggestion for line 58. https://codereview.appspot.com/324560043/diff/80001/cpm/Get.py File cpm/Get.py (right): https://codereview.appspot.com/324560043/diff/80001/cpm/Get.py#newcode58 cpm/Get.py:58: a = GetRequest(sys.argv[1], int(sys.argv[2])) Since this file will be a module (imported to CPM), please add: if __name__ == "__main__": GetRequest(sys.argv[1], int(sys.argv[2])) so that this last part of the code only executes if your file is executed as a script. More info: https://docs.python.org/2/tutorial/modules.html#executing-modules-as-scripts
Sign in to reply to this message.
Hi, I tried to make admission control logic on CPM_13 I put it on wrong folder RPM by mistake. BR, Vasileios Papageorgiou https://codereview.appspot.com/324560043/diff/80001/cpm/Get.py File cpm/Get.py (right): https://codereview.appspot.com/324560043/diff/80001/cpm/Get.py#newcode58 cpm/Get.py:58: a = GetRequest(sys.argv[1], int(sys.argv[2])) On 2017/10/12 12:18:08, Michail.Xirouchakis wrote: > Since this file will be a module (imported to CPM), please add: > > if __name__ == "__main__": > GetRequest(sys.argv[1], int(sys.argv[2])) > > so that this last part of the code only executes if your file is executed as a > script. More info: > > https://docs.python.org/2/tutorial/modules.html#executing-modules-as-scripts Acknowledged.
Sign in to reply to this message.
Hi, Younes me and Sikandar we defined some logic for admission control. Check the new version of cpm_13.py here https://codereview.appspot.com/324560043/patch/120001/130001 BR, Vasileios Papageorgiou
Sign in to reply to this message.
On 2017/11/30 15:56:04, Vasileios Papageorgiou wrote: > Hi, > > Younes me and Sikandar we defined some logic for admission control. > > Check the new version of cpm_13.py here > https://codereview.appspot.com/324560043/patch/120001/130001 > > BR, > > Vasileios Papageorgiou Hello, Regarding admission control... Where is the action to drop this packet as well as add rule to the OF switch? I don't see either. Or am I missing something?
Sign in to reply to this message.
On 2017/12/03 16:53:57, Michail.Xirouchakis wrote: > On 2017/11/30 15:56:04, Vasileios Papageorgiou wrote: > > Hi, > > > > Younes me and Sikandar we defined some logic for admission control. > > > > Check the new version of cpm_13.py here > > https://codereview.appspot.com/324560043/patch/120001/130001 > > > > BR, > > > > Vasileios Papageorgiou > > Hello, > > Regarding admission control... > > Where is the action to drop this packet as well as add rule to the OF switch? > I don't see either. > Or am I missing something? Hi Michael, Inside this else: # Check admission control logic self.admission_Control(datapath,in_port, pkt) would be returned a value of the allow[] Based on 1 or 0 it would allow or deny respectively. Though we have not called the _deny_flow(self, datapath, in_port, pkt_in, hard_timeout=1) yet so you are right but first we need to check that the logic is okay. I can add 2-3 lines to deny the flow is not so complicated BR, Vasileios Papageorgiou
Sign in to reply to this message.
On 2017/12/03 17:02:33, Vasileios Papageorgiou wrote: > On 2017/12/03 16:53:57, Michail.Xirouchakis wrote: > > On 2017/11/30 15:56:04, Vasileios Papageorgiou wrote: > > > Hi, > > > > > > Younes me and Sikandar we defined some logic for admission control. > > > > > > Check the new version of cpm_13.py here > > > https://codereview.appspot.com/324560043/patch/120001/130001 > > > > > > BR, > > > > > > Vasileios Papageorgiou > > > > Hello, > > > > Regarding admission control... > > > > Where is the action to drop this packet as well as add rule to the OF switch? > > I don't see either. > > Or am I missing something? > > Hi Michael, > > Inside this else: > > # Check admission control logic > self.admission_Control(datapath,in_port, pkt) > > would be returned a value of the allow[] > Based on 1 or 0 it would allow or deny respectively. > Though we have not called the _deny_flow(self, datapath, in_port, pkt_in, > hard_timeout=1) > yet so you are right but first we need to check that the logic is okay. > I can add 2-3 lines to deny the flow is not so complicated > > BR, > > Vasileios Papageorgiou Yes, please do it. And after you finalize the code, please run a few tests to check if your admission logic works in practice. And write the results in the weekly report.
Sign in to reply to this message.
Hi, I added the decision making function that is required after calling the admission conrol method. So , based on the calculations of Admission control function the flow would be denied or not. Check the code in here : https://codereview.appspot.com/324560043/patch/140001/150001 BR, Vasileios Papageorgiou
Sign in to reply to this message.
Hello Vasileios, I have not run your code. But it makes little sense to me that you would send multiple OF messages to the switch to deny this packet (and all consecutive packets of this flow). One OF message should be sufficient. I am referring to lines 528-532 of your commit: https://gits-15.sys.kth.se/vpap/Communication-System-Design_SDN/commit/aef85f... Moreover, you have commented out: #self._add_switch_flow(datapath, in_port, pkt, msg.buffer_id) (line 534) so how are admitted packets routed? BR, Michail
Sign in to reply to this message.
Hi, Check the update of Admission Control. https://codereview.appspot.com/324560043/diff/160001/RPM/cpm_13.py BR, Vasilis
Sign in to reply to this message.
https://codereview.appspot.com/324560043/diff/160001/RPM/cpm_13.py File RPM/cpm_13.py (right): https://codereview.appspot.com/324560043/diff/160001/RPM/cpm_13.py#newcode150 RPM/cpm_13.py:150: if packet_headers['tcp_dst'] or packet_headers['tcp_src'] == 10001: Why are they all at 100??? Doesn't make sens
Sign in to reply to this message.
Hello all, Code is currently correct and works as intended. Vasileios and I tested it in the classroom. Demo video: https://youtu.be/TNlpvK32svg Latest CPM with Admission Control: https://gits-15.sys.kth.se/vpap/Communication-System-Design_SDN/blob/cpm_dev_... BR, Michail
Sign in to reply to this message.
|
