Michael,  Are you using AAA Fastconnect allowing your controller to handle
radius requests instead of using a backend server?  While we haven't done
this ourselves, I know of others that have run into the same issue of not
being able to keep up with the auth requests.  You'll notice this even more
as smartphones will re-auth all the time.
- Don


On Fri, Jan 31, 2014 at 4:26 PM, Michael Hulko <[email protected]> wrote:

> One other wrench in this at least from the Aruba standpoint.... check the
> cpu load on the Auth process....  we found back in late October that one of
> our heaviest used controller (M3 running 6.1.3.7) was pegging over 90%
> utilization for the Auth process which at the time
> we believed the authentication was being additionally impacted (mostly
> drops).  It was indicated (source does not want to be mentioned) that there
> was a hard limit to the number of auth's per second the controller could
> handle (approx. 40 - 50/sec), at peak we were
> running around ~100/sec.  We upgraded to the version 6.3.x to resolve
> other issues.  We noticed that the system now spawned 3 Auth processes, but
> we still getting complaints.  We then discovered through TAC and internal
> investigation that a new dot1x throttling
> mechanism had been introduced in the version of code. This new
> "Throttling" was still impacting our authentications but saving the cpu's
> on the auth process.  We were instructed to adjust the Watermarks to reach
> a balance point from the defaults.  This is a slidiing scale....
> the higher the Watermarks, the higher the cpu process, but the less drops
> experienced.
>
> to view the cpu process:  "Show cpuload current | include auth"
>
> On 6.3x code:
>
> to view the Throttle parameters: "show dot1x counters"  (There is some
> math involved when the system decides to drop packets)
> to view the dropped auths: "show ap debug client-mgmt-counters" and look
> for the "Associations Dropped Due to Auth Throttling"
>
> In the end, the old addage still holds true..."You can never please 100%
> of the people 100% of the time....Keep calm and carry on"
>
> M
>
>
>
> On 2014-01-31, at 2:11 PM, Jeffrey Sessler wrote:
>
>  We noticed that the WLAN with band/load-steering enabled had a high
> report rate of Macintosh connectivity issues, and the WLAN that did not was
> trouble free.
>
> I suspect what was happening was this: Mac would initially associate
> (Ent-WPA2), then the controller would force it to move to another band
> and/or AP. It's at this point (a roam) that the Apple certificate issue
> would kick in, and it was hit or miss as to the Mac re-associating or
> failing. This was especially problematic when a Mac client was equidistant
> from two AP's.
>
> Turning off band/load steering pretty much eliminated the bulk of the
> connectivity issues, and trusting the certificate solved the rest.
>
> Band/load steering is just problematic because you can never predict how a
> client will react to it.
>
> Jeff
>
> >>> On Friday, January 31, 2014 at 10:57 AM, in message <
> CAPCnwUdh-=jawm78pjfuu1n9bhs9d_japthbfnwrrgsrbzg...@mail.gmail.com>,
> Norman Elton <[email protected]> wrote:
>    Interesting. What were the band-steering symptoms? Any way to pin the
> problem down to band-steering, or was it trial and error?
>
> Norman
>
>
> On Fri, Jan 31, 2014 at 1:44 PM, Edward Ip <[email protected]>wrote:
>
>>  I agree with Jeff, we recently disabled band steering on our Aruba
>> controllers and it has helped a bit.
>> *Edward Ip*
>> *Algonquin College* | 1385 Woodroffe Avenue | Room C316 | Ottawa |
>> Ontario | K2G 1V8 | Canada
>> algonquincollege.com
>>
>>   *From:* The EDUCAUSE Wireless Issues Constituent Group Listserv
>> [mailto:[email protected]] *On Behalf Of *Jeffrey
>> Sessler
>> *Sent:* Friday, January 31, 2014 1:40 PM
>>
>> *To:* [email protected]
>> *Subject:* Re: [WIRELESS-LAN] OS X 802.1x auth issue
>>
>>  We've seen the cert issue, and OS 10.8 and 10.9 don't seem to like
>> band/load-steering. The cert issue coupled with band-steering and/or
>> load-steering make the Mac's very unhappy.
>>    Jeff
>>
>> >>> On Friday, January 31, 2014 at 10:05 AM, in message <
>> CAPCnwUdAuZqKuFwOycKrGmXgiKCrb_Wy82=o5xc3be+o7an...@mail.gmail.com>,
>> Norman Elton <[email protected]> wrote:
>>    And a follow up. Has anyone actually confirmed that this bug is
>> actually causing client complaints? We do seem to riding a wave of
>> complaints from MacBook owners. We are only just now starting to
>> change cert trust settings. Hopefully we'll know more next week as
>> students have a chance to test things out over the weekend.
>>
>> Norman Elton
>> College of William & Mary
>>
>> On Fri, Jan 31, 2014 at 12:59 PM, Norman Elton <[email protected]>
>> wrote:
>> >> It also appears specific to certs based on 2048 bit keys. Also there
>> is no
>> >> cert validation delay upon initial connect... only when attempting to
>> >> reauth... ie after a death or a roam event.
>> >
>> > Can anyone confirm the bug only affects certs with 2048 bit keys? I
>> > don't see that listed anywhere in Apple's release. It's an interesting
>> > twist.
>> >
>> > Thanks!
>> >
>> > Norman Elton
>> > College of William & Mary
>>
>> **********
>> Participation and subscription information for this EDUCAUSE Constituent
>> Group discussion list can be found at 
>> http://www.educause.edu/groups/.<http://www.educause.edu/groups/>
>> ********** Participation and subscription information for this EDUCAUSE
>> Constituent Group discussion list can be found at
>> http://www.educause.edu/groups/.
>>  ********** Participation and subscription information for this EDUCAUSE
>> Constituent Group discussion list can be found at
>> http://www.educause.edu/groups/.
>>
>>
> ********** Participation and subscription information for this EDUCAUSE
> Constituent Group discussion list can be found at
> http://www.educause.edu/groups/.
>
> ********** Participation and subscription information for this EDUCAUSE
> Constituent Group discussion list can be found at
> http://www.educause.edu/groups/.
>
>
>
>
> Michael Hulko
> Network Analyst
>
> Western University Canada
> Network Operations Centre
> Information Technology Services
> 1393 Western Road, SSB 3300CC
> London, Ontario  N6G 1G9
>
> tel: 519-661-2111 x81390
> e-mail: [email protected] <mailto:[email protected] <[email protected]>>
>
>
>
>
>
> ********** Participation and subscription information for this EDUCAUSE
> Constituent Group discussion list can be found at
> http://www.educause.edu/groups/.
>
>

**********
Participation and subscription information for this EDUCAUSE Constituent Group 
discussion list can be found at http://www.educause.edu/groups/.

Reply via email to