[cisco-bba] 'sessions' versus 'PPPoA/E sessions' counts

ahmad fadly peditea at yahoo.com
Fri Jun 5 03:29:24 EDT 2009


Hi Frank,

Based on my experience if the router is high in cpu process there will be a lot of pending sessions because router cannot handle the call. It happened in ppp session using l2TP. Usually the session will be jammed in the router. That's why there are lot of pending number of session.

Using CAC perhaps can protect the router from processing a lot of number of call. It will reject the call when cpu reach certain value (depend on our setting) instead of accept the call and process it. I assumed your number of mean call per minutes is high because cpu proc. 

How many percent is your cpu proc?

I don't know if the problem will be the same as I had. Hope it can help you. 


Here is the sample problem I had:

C_10K#sh sss statistics 

Current Subscriber Statistics:

Number of sessions currently up: 70
Number of sessions currently pending: 928
Number of sessions currently authenticated: 70
Number of sessions currently unauthenticated: 928
Highest number of sessions ever up at one time: 10611
Mean up-time duration of sessions: 00:10:48
Total number of sessions up so far: 237972
Mean call rate per minute: 682, per hour: 47780
Number of sessions failed to come up: 175444
Access type based session count:
VPDN sessions = 998
Traffic-Class sessions = 140

C_10K#sh vpdn session state | i 00:00:
52591      1233       32679      -, -                 wiccn  00:00:09 0         
19592      3585       32679      -, -                 wiccn  00:00:05 0         
21825      209        32679      -, -                 wiccn  00:00:01 0         
49333      2774       32679      -, -                 wiccn  00:00:01 0         
6986       6068       32679      -, -                 wiccn  00:00:01 0         
20031      6810       32679      -, -                 wiccn  00:00:01 0         
44303      693        32679      -, -                 wiccn  00:00:01 0         
16123      4342       32679      -, -                 wiccn  00:00:01 0         
11714      3903       32679      -, -                 wiccn  00:00:09 0         
51635      9267       32679      -, -                 wiccn  00:00:09 0         
59529      7723       32679      -, -                 wiccn  00:00:09 0         
5228       10439      32679      -, -                 wiccn  00:00:09 0         
55820      2991       32679      -, -                 wiccn  00:00:09 0         
49586      3899       32679      -, -                 wiccn  00:00:01 0         

---

C_10K#sh proc cpu his
                                                              
    44455555333333333344444555551111                         4
    11111111777773333333333999994444     33333     77777111111
100                                                           
 90                                                           
 80                                                           
 70                                                           
 60                        *****                              
 50    *****               *****                              
 40 *************     **********                             *
 30 ****************************                             *
 20 ****************************                             *
 10 ********************************               *****     *
   0....5....1....1....2....2....3....3....4....4....5....5....
             0    5    0    5    0    5    0    5    0    5    
               CPU% per second (last 60 seconds)
       11                                                     
    8380034444456434344644473346745547354475666544478999999999
    2730013159128490980609559821084218756790224893684999999999
100    #*                                            #########
 90    #*                                            #########
 80 * *#*                  *         *    *        **#########
 70 * *##       *      *   *    *    *    *        **#########
 60 * *##       *      *   *   **    * *  * ****   **#########
 50 * *##   ** **    * * ***   ***** * ********** **##########
 40 #**## ******#***************#*********#*#*##**############
 30 #*###***##################################################
 20 ##########################################################
 10 ##########################################################
   0....5....1....1....2....2....3....3....4....4....5....5....
             0    5    0    5    0    5    0    5    0    5    
               CPU% per minute (last 60 minutes)
              * = maximum CPU%   # = average CPU%


--

Fadly


--- On Fri, 6/5/09, Frank Bulk <frnkblk at iname.com> wrote:

> From: Frank Bulk <frnkblk at iname.com>
> Subject: RE: 'sessions' versus 'PPPoA/E sessions' counts
> To: "'ahmad fadly'" <peditea at yahoo.com>, cisco-bba at puck.nether.net, frnkblk at iname.com
> Date: Friday, June 5, 2009, 10:33 AM
> Thanks, but that doesn't solve the
> underlying issue of why I'm seeing so
> many call attempts.
> 
> Frank
> 
> -----Original Message-----
> From: ahmad fadly [mailto:peditea at yahoo.com]
> 
> Sent: Thursday, June 04, 2009 8:43 PM
> To: cisco-bba at puck.nether.net;
> frnkblk at iname.com
> Subject: Re: 'sessions' versus 'PPPoA/E sessions' counts
> 
> 
> I seem too many pending and fail number of sessions. If you
> see your cpu
> proc is to high and your box is overly busy, try to use CAC
> (call admission
> control). It works to prevent overloading cpu and protect
> the router
> resources. 
> 
> You can use sh call admission statistics to see the CAC
> statistic
> C10K#sh call admission statistics 
> Cac New Model (SRSM) is ACTIVE
> Total call Session charges: 0, limit 90
> Total calls rejected 1669, accepted 340854
> Reject reason: CPU-limit: 1669 SessionCharges 0
> Current actual CPU: 14%, Limit: 90%
> Hardware CAC is currently not active
> 
> -----
> Fadly
> 
> > Date: Thu, 04 Jun 2009 16:11:51 +0300
> > From: Tassos Chatzithomaoglou <achatz at forthnet.gr>
> > To: cisco-bba at puck.nether.net
> > Subject: Re: [cisco-bba] 'sessions' versus 'PPPoA/E
> > sessions' counts
> > 
> > Number of sessions currently up + Number of sessions
> > currently pending = PPPoA + PPPoE
> > 
> > PPPoA or PPPoE doesn't take into account whether the
> > sessions have finished their negotiation.
> > 
> > --
> > Tassos
> > 
> > Frank Bulk wrote on 04/06/2009 13:22:
> > > When I looked at my "sh subscriber statistics" I
> > noticed that 
> > > - my "sessions currently up" does not add up to
> the
> > number of PPPoA and
> > > PPPoE sessions: how can that be?
> > > - I have a very high "Number of sessions failed
> to
> > come up"
> > > - my call rate is 3341/minute, which is
> 56/second
> > > 
> > > I know I have lots of provisioned VP/VCs that
> don't
> > have an active DSLAM
> > > port, but I'm not sure if that contributes to an
> > explanation.
> > > 
> > > If there's a debug I need to turn on to "see",
> please
> > suggest the
> > > appropriate one.  It seems that the box is
> overly
> > busy.
> > > 
> > > Frank
> > > 
> > >
> >
> ==================================================================
> > > Router#sh subscriber statistics
> > > 
> > > Current Subscriber Statistics:
> > > 
> > > Number of sessions currently up: 2596
> > > Number of sessions currently pending: 1327
> > > Number of sessions currently authenticated: 2596
> > > Number of sessions currently unauthenticated:
> 1327
> > > Highest number of sessions ever up at one time:
> 2647
> > > Mean up-time duration of sessions: 00:01:08
> > > Total number of sessions up so far: 543761664
> > > Mean call rate per minute: 3341, per hour:
> 200502
> > > Number of sessions failed to come up: 543534572
> > > Access type based session count:
> > > PPPoA sessions = 3711
> > > PPPoE sessions = 212
> > > 
> 
> 


      


More information about the cisco-bba mailing list