cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
557
Views
0
Helpful
3
Replies

High number of output drops

toprock1970
Level 1
Level 1

I have a number of stacked 3750E switches that all have high numbers of output drops on all of the interfaces that are up:

Switch Ports Model              SW Version            SW Image
------ ----- -----              ----------            ----------
*    1 54    WS-C3750E-48PD     15.0(1)SE2            C3750E-UNIVERSALK9-M
     2 54    WS-C3750E-48PD     15.0(1)SE2            C3750E-UNIVERSALK9-M
     3 54    WS-C3750E-48PD     15.0(1)SE2            C3750E-UNIVERSALK9-M
     4 54    WS-C3750E-48PD     15.0(1)SE2            C3750E-UNIVERSALK9-M
     5 54    WS-C3750E-48PD     15.0(1)SE2            C3750E-UNIVERSALK9-M
     6 54    WS-C3750E-48PD     15.0(1)SE2            C3750E-UNIVERSALK9-M
     7 54    WS-C3750E-48PD     15.0(1)SE2            C3750E-UNIVERSALK9-M
     8 54    WS-C3750E-48PD     15.0(1)SE2            C3750E-UNIVERSALK9-M
     9 54    WS-C3750E-48PD     15.0(1)SE2            C3750E-UNIVERSALK9-M
 

#sh int | inc drops:.[1-9]
  Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 3091143
  Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 1888390
  Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 1795
  Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 3319808
  Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 8620
  Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 74405
  Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 20973
  Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 1048535
  Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 741080
  Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 4271
 

I cleared the interface statistics yesterday.

When I look  at my other stacked switches (WS-C3750G-24PS) they don't have any drops.

Is this a common problem on the WS-C3750E-48PD? Is there  known fix?
Does anyone have any advise one what I should look at to try and get to the bottom of this - maybe qos or something else?

 

Any help would be gretaly appreciated
 

Regards
Jason

3 Replies 3

Joseph W. Doherty
Hall of Fame
Hall of Fame

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

Often enabling QoS, using defaults, on any 2960/3560/3750 will cause additional drops.

Even without QoS, the 2960/3560/3750 series don't have a lot of buffer space and are all prone to transient congestion drops.  One possible reason, you might see more drops on a 3750E vs. a 3750G, the former might have a 10g uplink, the latter having 10g is very rare.  If you have a 10g link, it can more easily overrun a gig egress port.

Can drops be mitigated?  Often they can.  The 3750s have 2 MB of RAM for each bank of 24 edge ports and for the module ports.  Moving your busy ports onto different banks might help.  Also, buffer tuning, when QoS is enabled, might reduce drops too. (I've had good success with more of shared pool approach [more like the 3550s] rather than a reserved per interface approach, for reducing drops.)

Thanks for the response Joseph.

We do have 10gig uplinks so you could very well be onto something there.

I'm not too hot on Qos so I'll take a look into it.
What I'm not ggetting my head around is why the drops are happening when my monitoring (Solarwinds) indicates that the links are nowhere near getting maxed out. I'd expect to see high traffic on some of the links going to my VM Hosts  if the 10gig links were trying to push too much traffic to a particular host.

 

Regards
Jason

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

This might help: http://notalwaysthenetwork.com/2014/01/06/microburst-detection-with-wireshark/

Review Cisco Networking for a $25 gift card