Guest

Preview Tool

Cisco Bug: CSCuo13237 - TX buffer size was not updated to the max 4K due to non-updated ISO

Last Modified

Apr 06, 2015

Products (1)

  • Cisco Videoscape Distribution Suite Transparent Caching

Known Affected Releases

5.1.1

Description (partial)

Symptom:
Issue: ENIC driver generating Error Messages showing in Kernel and eventually leading to kernel freeze
ce-1 kernel: enic 0000:0a:00.0: eth4: BUG! Tx ring full when queue awake!
ce-1 last message repeated 785 times

Observing the /var/log/messages on the various cache engines, the ENIC driver flagged around 0.5M of this bug messages during 5-7 hours. These messages were seen on both the data NICs (standard packets) and the iSCSI NICs (jumbo packets) on the cache engines.
When the ENIC driver stops the TX queue, the packet in queue are not transmitted out and the cache defers these packets for later re-transmission. When this is done, CPU resources are allocated until the packet is transmitted. A buildup of CPU resource allocation, in addition to the hundreds of thousands of ENIC TX queue stops and "BUG..." messages flagged ? can eventually lead to a very busy CPU and system freeze. The system freeze is detected by the built-in watchdog ? that restarts the machine. The WD intervention is per design, in case of such platform overload.
In addition to the above, and due to the use of an older ISO image compared to the GA, the ENIC size of the TX ring buffers was not properly set to the required value.

Conditions:
System running with load
Bug details contain sensitive information and therefore require a Cisco.com account to be viewed.

Bug Details Include

  • Full Description (including symptoms, conditions and workarounds)
  • Status
  • Severity
  • Known Fixed Releases
  • Related Community Discussions
  • Number of Related Support Cases
Bug information is viewable for customers and partners who have a service contract. Registered users can view up to 200 bugs per month without a service contract.