Guest

Preview Tool

Cisco Bug: CSCtc88016 - Rare race condition between dbmon and repl scripts breaks replication

Last Modified

Nov 03, 2012

Products (1)

  • Cisco Unified Communications Manager (CallManager)

Known Affected Releases

8.0

Description (partial)

Symptom:
When setting replication, dbmon sets the replstate rtmt counter to 1. It then checks if replication is indeed being setup for its node by looking up the node name in publisher's nodelist variable.

 If DbMon has fired the script to setup replication and publisher's nodelist doesn't have its node name in it, then DbMon takes it as a hung call (hung for whatever reason, firewall, network problem). It tries to kill the hung call and redo the steps of replication setup. 

In this particular case the race condition happened within a span of 20 seconds when the publisher had completed the setup for this subscriber hence the nodename was not there in publisher's nodelist but publisher hadn't notified the subscriber dbmon that replication setup is completed. DbMon on the sub took it as a hung script call and tried to recover when it shouldn't have done so.

Need to tighten the check for nodename in nodelist so that this slip doesn't happen.

Conditions:

Extremly rare race condition
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.