Cisco Blog » The Platform

2015年10月1日星期四

Cisco Notification Alert -3850 Series Switch-01-Oct-2015 16:50 GMT

 

 

 

 

 

 

 


Known Bugs - Catalyst 3850 Series Switches

Alert Type:
Updated *
Bug Id:
CSCuu71450
Title:
Beni-E2:Client traffic fails after MA to MC roaming
Status:
Fixed
Severity:
2 Severe
Description:

Symptom:
Client traffic fails after MA to MC roaming

Conditions:
Client traffic fails after roaming from MA to MC. Client lost the IP after roaming though Anchor-Foreign relationship established fine

Workaround:
None

Further Problem Description:

Last Modified:
03-SEP-2015
Known Affected Releases:
0.6(0), 10.2(121.1), 16.1(0), 40.1(100)DSP
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E2, 3.7(2)E
Alert Type:
New
Bug Id:
CSCur17429
Title:
SNMP: IEEE8023-LAG-MIB does not work on the switch.
Status:
Fixed
Severity:
2 Severe
Description:

IEEE8023-LAG-MIB is not working on 3850

Symptom:
The query IEEE8023-LAG-MIB cannot be executed on the switch.

Conditions:
The issue occurs when you have port channel configured on the switch.

Workaround:
The workaround is to use CISCO-LAG-MIB.

Further Problem Description:

Last Modified:
09-SEP-2015
Known Affected Releases:
15.2(3.2.23)E
Known Fixed Releases:
15.2(2)E2, 15.2(3)E1, 15.2(4.0)ST, 15.2(4.0.64a)E, 15.2(5.0)ST, 3.6(2)E, 3.7(1)E
Alert Type:
Updated *
Bug Id:
CSCus99367
Title:
3850 re-writes mobility capwap data packets with TTL 9
Status:
Fixed
Severity:
2 Severe
Description:

Symptom:
3850 re-writes mobility capwap data packets with TTL 9

Conditions:
3850 re-writes mobility capwap data packets with TTL 9. As a result any capwap data packets(DHCP/ARP etc) sent via mobility tunnel will have a TTL set to 9 which means these packets will not be able to make more than 9 hops.

In an environment where 3850 wlan is anchored to another WLC/Switch >9 hops away, clients will not be able to get DHCP IP.

Hardware affected: 3850/3650
Software versions impacted: 3.3.x, 3.6.0, 3.6.1, 3.6.2, 3.7.0, 3.7.1

5760 is not impacted by this issue

Workaround:
Shorten the path between the foreign and anchor or upgrade to fixed image(IOS-xe 3.7.2 or 3.6.3).

Further Problem Description:
None.

Last Modified:
04-SEP-2015
Known Affected Releases:
15.2(3)E
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E2, 3.6(3)E, 3.7(2)E
Alert Type:
Updated *
Bug Id:
CSCuu97048
Title:
Traffic is dropped due to static mac entry on foreign interface
Status:
Fixed
Severity:
2 Severe
Description:

Symptom:
Traffic sourced from particular MAC address is getting dropped when this MAC address is statically added for another physical interface in the same broadcast domain.

Conditions:
There is static MAC address entry which is pointing for particular interface. Traffic sourced from this MAC address is silently dropped on other interfaces.

Workaround:
Remove static MAC address entry

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
15.2(3)E
Known Fixed Releases: *
15.2(2)E3, 3.6(3)E
Alert Type:
Updated *
Bug Id:
CSCut26365
Title:
Packet drop on 3850 by an unrelated ACL entry
Status:
Fixed
Severity:
2 Severe
Description:

Symptom:
TCP ack packet is discarded by unrelated ACE on 3850.

Conditions:
"establish" option is used.

Workaround:
The problem happened with tcp flags which have multiple bit set. If customer cofig as separate bit, then there is no issue.

The ace: deny tcp any any eq ftp established

Configured as two aces (because the tcp flag establish means "ack or rst"

deny tcp any any eq ftp ack
deny tcp any any eq ftp rst

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
3.7(0)E
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E2, 3.6(3)E, 3.7(2)E
Alert Type:
Updated *
Bug Id:
CSCuu31131
Title:
Memory leak under *MallocLite* / tdl code for ipv4 and l3m
Status:
Fixed
Severity:
2 Severe
Description:

Symptom:
Memory leak is seen in *MallocLite* process. if MallocLite is disabled, we see the increase in memory under the TDL code for l3m and ipv4. Further investigation is going on to identify the source code involved in triggering the leak.

Conditions:
Not known

Workaround:
Not known

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
3.6(2)E
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E2, 15.2(4.0.64a)E, 15.2(4.0.8)E, 15.2(5.0)ST, 3.6(3)E, 3.7(2)E
Alert Type:
Updated *
Bug Id:
CSCuu70556
Title:
stack manager crash @ dplr_pd_update_ring_status_chg
Status:
Fixed
Severity:
2 Severe
Description:

Symptom:
3850 switch running 3.7.1.E might experience a stack manager crash.

Conditions:
unknown yet

Workaround:
unknown yet

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
15.2(3.7.1E)
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E2, 3.6(3)E, 3.7(2)E
Alert Type:
Updated *
Bug Id:
CSCut44425
Title:
SFP remove and crash when "show platform port-asic 0 read counters"
Status:
Fixed
Severity:
2 Severe
Description:

Symptom:
SFP remove and crash occur when run "show platform port-asic 0 read counters" command

Conditions:
when "show platform port-asic 0 read counters" command is run

Workaround:
We can use the following command as workaround now.

Switch#show platform fwd-asic counters tla ?
AQM Active Queue Management
ASE Acl Search Engine
EGR Egress Global Resolution
EPF Egress Scheduler Module
EQC Egress Scheduler Module
ESM Egress Queue Controller
FPE Flexible Parser
FPS Flexible Pipe Stage
FSE Fib Search Engine
IGR Ingress Global Resolution
IPF Ingress Port FIFO
IQS Ingress Queues and Scheduler
NFL Netflow
NIF Network Interface
PBC Packet Buffer Complex
PIM Protocol Independent Multicast
PLC Policer
RMU Recirculation Multiplexer Unit
RRE Reassembly Engine
RWE Rewrite Engine
SEC Security Engine
SIF Stack Interface
SQS Stack Queues And Scheduler
SUP Supervisor Interface

Switch#show platform fwd-asic counters tla NIF detail asic 0 sw 4
Starting with asic 0

NifRxByteDestinationGroupStats on Asic 0
[0]
rxUnicastBytes1 0x00000000
rxUnicastBytes0 0x00000184
rxMulticastBytes1 0x00000000
rxMulticastBytes0 0x00000000
rxBroadcastBytes1 0x00000000
rxBroadcastBytes0 0x00000000

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
15.5(2)T
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E2, 3.6(3)E, 3.7(2)E
Alert Type:
Updated *
Bug Id:
CSCuu87436
Title:
3850 / 3.7.1E / "switchport block multicast" blocks multicast ip traffic
Status:
Fixed
Severity:
2 Severe
Description:

Symptom:
+ Multicast ip traffic blocked
+ Eg: MDNS traffic

Conditions:
+ Requires the command "switchport block multicast" configured on the egress interfaces.

Workaround:
+ Remove "switchport block multicast"

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
3.6(0)E
Known Fixed Releases: *
15.2(2)E3, 3.6(3)E
Alert Type:
Updated *
Bug Id:
CSCur16497
Title:
Egress WCCP does not work when configured with Egress SPAN
Status:
Fixed
Severity:
2 Severe
Description:

Symptom:
On a stack of 3850 switches (WS-C3850-48T-E) running 3.3.4SE, when wccp redirect out is configured on the vlan interface the traffic is not getting redirected

Conditions:
"ip wccp 61 redirect out" or "ip wccp 62 redirect out" configured on the WAN side along with egress wccp span.

Workaround:
remove egress span from the same interface as egress wccp.

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
15.0(1)EZ4
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E1, 3.6(3)E, 3.7(1)E
Alert Type:
Updated *
Bug Id:
CSCuu86656
Title:
2 different policies with same order of class-maps on same asic
Status:
Fixed
Severity:
2 Severe
Description:

Symptom:

traffic hits in correct class-map if we configure 2 different policies with same order of class-maps(these class-maps are port-range based acls) on same asic.
Conditions:when the policy contains L4 range match (e.g, tcp port > 100), and only reboot.
Workaround:manually reconfig after reboot.
More Info:












Last Modified:
03-SEP-2015
Known Affected Releases:
Rev
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E2, 3.7(2)E
Alert Type:
Updated *
Bug Id:
CSCut22611
Title:
3850 - 1 GiG 1000BaseLX negotiates to "HALF DUPLEX" after reload
Status:
Fixed
Severity:
2 Severe
Description:

Symptom:
1 Gig port (1000BaseLX) sometime starts operating at HALF DUPLEX mode after a reload.

Shut/no shut or plugging/unplugging the SFP does not fix the issue. Only way to fix it, is remove the SFP, hardcode the duplex to full and insert SFP back.

Conditions:
WS-C3850-24U running 3.3.5
Both ports need to be connected to C3850-NM-4-1G.
GBIC used- GLC-LH-SMD for 1000BaseLX

Workaround:
Not Available yet.

UDLD should be kicked in to stop any possible layer 2 loop due to this unidirectional situation but this is anyway not a valid workaround to stop the SFP port operating at HALF DUPLEX.

Further Problem Description:

Last Modified:
03-SEP-2015
Known Affected Releases:
15.0(1)EZ5
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E1, 15.2(3)E2, 3.7(1)E, 3.7(2)E
Alert Type:
Updated *
Bug Id:
CSCuv69263
Title:
3850 crashes while running 'no ip pim rp-address'
Status: *
Other
Severity: *
3 Moderate
Description:

Symptom:
A 3850 may crash while running the command 'no ip pim rp-address'

Conditions:
Running the command 'no ip pim rp-address'

Workaround:
None at this time.

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
15.0(1)EX2
Known Fixed Releases:
Alert Type:
Updated *
Bug Id:
CSCuw02869
Title:
%PLATFORM_PM-6-MODULE_REMOVED msg upon multiple reset of the SFPs
Status: *
Other
Severity: *
3 Moderate
Description:

Symptom:
Below error msg seen on the reset of the SFPs.

*Aug 31 02:41:54.217: %PM-4-ERR_DISABLE: gbic-invalid error detected on Te3/1/3, putting Te3/1/3 in err-disable state
*Aug 31 02:41:54.221: %PLATFORM_PM-6-MODULE_REMOVED: SFP module with interface name Te3/1/3 removed

Conditions:
-NA-

Workaround:
-NA-

Further Problem Description:

Last Modified:
01-OCT-2015
Known Affected Releases:
15.2(3.7.2)
Known Fixed Releases:
Alert Type:
Updated *
Bug Id:
CSCut06428
Title:
backup flexlink with multicast fast-convergence is leaking igmp leave
Status:
Fixed
Severity:
3 Moderate
Description:

Symptom:
Backup flexlink port with "multicast fast-convergence" is leaking igmp leave packets causing loop

Conditions:
NG3k switch with flexlink configuration and "multicast fast-convergence" option

Workaround:
disable "multicast fast-convergence" option

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
3.6(0)
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E2, 15.2(4.0.64a)E, 15.2(5.0)ST, 3.6(3)E, 3.7(2)E
Alert Type:
Updated *
Bug Id:
CSCur94225
Title:
Output drop doesn't count up
Status:
Fixed
Severity:
3 Moderate
Description:

Symptom:
on C3850, "Total output drops" has no counter even though the drop is happening.

Conditions:
output drop happens.

Workaround:
none

Further Problem Description:

Last Modified:
02-SEP-2015
Known Affected Releases:
15.2(2)E1, 3.6(1)E
Known Fixed Releases: *
15.2(1)SY1, 15.2(2)E3, 15.2(3)E1, 15.2(4.0.64a)E, 15.2(5.0)ST, 3.7(1)E
Alert Type:
Updated *
Bug Id:
CSCut29532
Title:
FlexLink and QinQ breaks communication on 3850
Status:
Fixed
Severity:
3 Moderate
Description:

Symptom:
- on dot1q-tunnel interface in egress direction the outer tag (Service Provider's VLAN ID) is not stripped when forwarding traffic to Customer's equipment

Conditions:
- 3850 with 03.06.00 or 03.06.01 or 03.07.00
- QinQ is configured
- FlexLink is configured on dot1q-tunnel interface

!
interface GigabitEthernet1/0/1
switchport access vlan 1082
switchport mode dot1q-tunnel
switchport backup interface Gi1/0/3 <================= FlexLink configured
l2protocol-tunnel cdp
l2protocol-tunnel stp
no cdp enable
!

Workaround:
- remove FlexLink configuration from dot1q-tunnel interface

!
interface GigabitEthernet1/0/1
switchport access vlan 1082
switchport mode dot1q-tunnel
! switchport backup interface Gi1/0/3 <================= FlexLink configuration is commented out (removed)
l2protocol-tunnel cdp
l2protocol-tunnel stp
no cdp enable
!

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
15.2(3)E
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E2, 15.2(4.0.64a)E, 15.2(5.0)ST, 3.6(3)E, 3.7(2)E
Alert Type:
Updated *
Bug Id:
CSCup47260
Title:
3850 FED crash on member switch with %PDSLIB-3-OBJFREE
Status:
Fixed
Severity:
3 Moderate
Description:

Symptom:
A 3850 member switch will experience a FED crash:
%PDSLIB-3-OBJFREE: 1 fed: A PACKET object at 0x8100 has already been freed

Conditions:
It is suspected that this was triggered in a corner-case timing event in how ICMP redirects are handled in the CPU queue. Another trigger condition that is required is when "debug pds pktbuf-last" is enabled.

Note: The output from "show debug" will NOT reflect whether any pds debugs are enabled.

Workaround:
Disable debugging for the pds queue

Further Problem Description:
-The crash was seen with a large "RX unsuspend send count" from the output of "show platform punt statistics port-asic 0 cpuq6 direction rx".

-There were also reported MACFLAPS occurring which may/may not tie into the crash itself. To assess the rate of which mac flaps are happening, "show platform matm counters" will show move_event/fed_learn_event counters incrementing.

Last Modified:
04-SEP-2015
Known Affected Releases:
15.0(1)EZ1
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E, 3.6(3)E, 3.7(0)E
Alert Type:
Updated *
Bug Id:
CSCus30573
Title:
NG3k switch do not clear inactive mac addresses
Status:
Fixed
Severity:
3 Moderate
Description:

Symptom:
NG3k switch do not clear inactive mac addresses when port-security and "aging type inactivity" configured.

Conditions:
NG3K with port-security/aging type inactivity and end device is switching in between trunk port and phone

Workaround:
issue "clear port-security all"

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
15.0(1.99)
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E1, 3.6(3)E, 3.7(1)E
Alert Type:
Updated *
Bug Id:
CSCut21614
Title:
SR632757933:Unable to ping beyond 3850 due to same vlan id & l3ifindex
Status:
Fixed
Severity:
3 Moderate
Description:

Symptom:
L2/L3 connectivity can go down when another port in a different vlan is shut/no shut.

Conditions:
The internal l3ifindex and the vlan id share the same index value.

Workaround:
Assign a different vlan id to the port so that the l3ifindex no longer shares the same index with the vlan.

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
15.2(2.0.0)
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E2, 3.6(3)E, 3.7(2)E
Alert Type:
Updated *
Bug Id:
CSCut49440
Title:
3850 class-map "match" doesn't work correctly
Status:
Fixed
Severity:
3 Moderate
Description:

Symptom:
If class-map match-any has "match dscp" and "match ip precedence" as follows, traffic with dscp value doesn't hit the policy-map.

!
class-map match-any DSCP-CS5
match dscp cs5
match ip precedence 5
!
policy-map PM-PQ
class DSCP-CS5
police cir 8000 conform-action transmit exceed-action drop

Conditions:
"match ip precedence " is configured in same the class-map.

Workaround:
Either removing " match ip precedence x" or configuring "match dscp xx" after "match ip precedence x" would resolve the issue.

!
class-map match-any DSCP-CS5
match ip precedence 5
match dscp cs5

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
3.7(0)E
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E2, 3.6(3)E, 3.7(2)E
Alert Type:
Updated *
Bug Id:
CSCuu36466
Title:
after using "no mdix auto" command the link status is incorrect on 3850
Status:
Fixed
Severity:
3 Moderate
Description:

Symptom:
after using "no mdix auto" command the link status is incorrect on 3850

Conditions:
- disable "mdix auto" on remote side switch Cat2960S
- Cat3850 connect to Cat2960S with straight cable (use shut/no shut in repro)
- disable "mdix auto" on 3850

Workaround:
- use corss cable
- enable "mdix auto"
- reconnect the cable(or shut/no shut the interface)

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
15.2(2)E2
Known Fixed Releases: *
15.2(2)E3, 3.6(3)E
Alert Type:
Updated *
Bug Id:
CSCuu10251
Title:
CMI show CLI crash when system runs low in memory
Status:
Fixed
Severity:
3 Moderate
Description:

Symptom:
System crash while issuing CMI show CLI

Conditions:
When the system runs low in memory, it will crash while issuing show CMI.

Workaround:
NA

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
3.6.0
Known Fixed Releases: *
15.2(2)E3, 3.6(3)E
Alert Type:
Updated *
Bug Id:
CSCuu29813
Title:
DHCP snoop on uplink vlan create WCDB error, does not match binding vlan
Status:
Fixed
Severity:
3 Moderate
Description:

Symptom:
wireless clients will not get an IP.

Conditions:
DHCP snooping on uplink vlan. 3850 acting as a router. DHCP response coming back on a different vlan than client vlan.

Seen after upgrading to 03.06.02.

Workaround:
remove DHCP snooping from uplink vlan.

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
n/a
Known Fixed Releases: *
15.2(2)E3, 15.2(4.0.29)E, 15.2(4.0.64a)E, 15.2(5.0)ST, 3.6(3)E
Alert Type:
Updated *
Bug Id:
CSCut68387
Title:
FED Memory corruption Crash due to CGM
Status:
Fixed
Severity:
3 Moderate
Description:

Symptom:
FED process crash with errors mentioning memory corruption

Conditions:
Unknown

Workaround:
Unknown

Further Problem Description:

Last Modified:
03-SEP-2015
Known Affected Releases:
15.2(3.16)M0.7
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E2, 3.7(2)E
Alert Type:
Updated *
Bug Id:
CSCuu56466
Title:
"Total output drops" counter of a certain ports does not increment
Status:
Fixed
Severity:
3 Moderate
Description:

Symptom:
"Total output drops" counter for Giga1/0/1 and Giga1/0/25 does not work.

Conditions:
Cat 3850/Cat 3650

Workaround:
None

Further Problem Description:

Last Modified:
04-SEP-2015
Known Affected Releases:
15.2(3)E
Known Fixed Releases: *
15.2(2)E3, 3.6(3)E
Alert Type:
Updated *
Bug Id:
CSCuu34717
Title:
3850 cts enforcement for multicast traffic
Status:
Fixed
Severity:
4 Minor
Description:

Symptom:
3850 platform configured for trustsec will check policies for multicast traffic (we do have enforcement for multicast traffic). Multicast traffic is classified to (sgt,0) if one is defined, otherwise hits the default sgacl (0,0).

The same behavior is also for 3650 and 5760.

Most of the other platform is not using trustsec enforcement for multicast traffic.

This is platform specific behavior and bug has been created to document it.

Conditions:
cts on 3850/3650/5760 and muticast traffic

Workaround:
If the catchall is defined as a permit instead of deny, then multicast traffic is forwarded.

Further Problem Description:
The fix involves ensuring the deny acl is not applied to multicast traffic.

Last Modified:
04-SEP-2015
Known Affected Releases:
3.7(0)
Known Fixed Releases: *
15.2(2)E3, 15.2(3)E2, 3.6(3)E, 3.7(2)E
Alert Type:
Updated *
Bug Id:
CSCun63989
Title:
Express setup logs (express_setup.debug) should include more details
Status:
Fixed
Severity:
4 Minor
Description:

Symptom:
There is no way to tell which switch triggered the reset when the Mode button is pressed.

Conditions:
3850 switch with a Mode button.

Workaround:
None

Further Problem Description:
This enhancement will include for information in the express setup file regarding the switch number and timestamp.

Last Modified:
04-SEP-2015
Known Affected Releases:
15.0(1)EZ2
Known Fixed Releases: *
15.2(1)SY1, 15.2(2)E3, 15.2(3)E1, 15.2(4.0)ST, 15.2(4.0.64a)E, 3.6(3)E, 3.7(1)E
Alert Type:
New
Bug Id:
CSCuw09562
Title:
stale config with interface templates
Status:
Open
Severity:
5 Cosmetic
Description:

Symptom:
"spanning-tree port fast" config was in template. After adding template using "source template ", the config was seen out of template

Conditions:
By configuring "source template "

Workaround:

Further Problem Description:

Last Modified:
03-SEP-2015
Known Affected Releases:
41.2(0)DSP
Known Fixed Releases:

Find additional information in Bug Search index.

 

2013 Cisco and/or its affiliates. All rights reserved. Terms & Conditions | Privacy Statement | Cookie Policy | Trademarks

 

没有评论:

发表评论