Preview Tool

Cisco Bug: CSCtf19999 - TCP process can crash when BGP sessions flap

Last Modified

May 16, 2018

Products (11)

  • Cisco Carrier Routing System
  • Cisco CRS-1 Line Card Chassis (Dual)
  • Cisco CRS-1 16-Slot Line Card Chassis
  • Cisco CRS-1 Line Card Chassis (Multi)
  • Cisco IOS XR Software
  • Cisco CRS-1 4-Slot Single-Shelf System
  • Cisco CRS-1 Fabric Card Chassis
  • Cisco CRS-1 8-Slot Line Card Chassis
  • Cisco CRS-1 8-Slot Single-Shelf System
  • Cisco CRS-1 Multishelf System
View all products in Bug Search Tool Login Required

Known Affected Releases


Description (partial)

On an IOS-XR router running 3.8.1 the TCP process may crash if BGP/TCP sessions are flapped at a high rate. This problem is closey related to CSCtd20027.
During the testing of the fix and SMU for CSCtd20027 it was discovered that if BGP and TCP sessions are flapped at a high enough rate the TCP process may still crash even with the CSCtd20027 SMU loaded.

TCP has a feature to dump the state of a terminating session to a file.  The file is typically used for post-mortem analysis. We keep only the most recent 2000 such files in an in-memory file system . They are compressed binary files that take up approx 1KB space for file. However, the actual work to dump the state is offloaded to a worker thread by making a copy of the terminating session. In this particular case, the worker thread was unable to keep up with the number of dump copies that were queued to it - hence the increase in TCP's memory and subsequent crash.
Bug details contain sensitive information and therefore require a 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.