I have a 3750x stack running 15.0(2)SE6.
Today I started seeing some weird problems in the logs
" %ADJ-3-RESOLVE_REQ : Adj resolve request: Failed to resolve <IP Address> <vlan ID>..
To my knowledge no changes have been made except updating the NTP and Timezone settings of the switch.
The problems is that before I rebooted a vmWare - Host all communication with said host worked like a champ. After reboot of said host, the only communications that work are from the host outward. Anyone have any ideas? I would love to reboot, but as this is production hours that is not an option.
In a Stack 3850
Base Ethernet MAC Address : f8:a5:c5:06:76:00
Motherboard Assembly Number : 73-15800-08
Motherboard Serial Number : FOC20484XUZ
Model Revision Number : AB0
Motherboard Revision Number : A0
Model Number : WS-C3850-48P
System Serial Number : FOC2048U14U
Switch Ports Model SW Version SW Image Mode
------ ----- ----- ---------- ---------- ----
* 1 56 WS-C3850-48P 03.07.03E cat3k_caa-universalk9 INSTALL
2 56 WS-C3850-48P 03.07.03E cat3k_caa-universalk9 INSTALL
Seems that if your logged in you can't search this bug. If you google it while your logged out google will come up with a hit and you can read the bug.
The the text under conditions does not make any sense to me. Nobody wants to see a log message every time the device needs to resolve an arp entry. You shouldn't have to disable something the makes the switch perform better just to avoid filling up the log with garbage.
The message interpreter tells you
Explanation An internal software error occurred.
The details tell you to open a tac case.
Bottom line is..Will it be fixed, and in what code level will it be fixed.
If we are talking about 15.0(2)SE6 on 3750x you should try "no ip cef optimize neighbor resolution" - that should do the trick. (this option was disabled by default on 12.2 train, but enabled by default on 15.0) The other way is to try to hunt down all arp "incomplete" entries and check the issues with them - that would solve the problem with messages too.