Project

General

Profile

Actions

Bug #4494

closed

Accumulation of cf-agent processes due to locking on CFEngine tcdb lock file

Added by Nicolas CHARLES about 10 years ago. Updated about 9 years ago.

Status:
Released
Priority:
1
Category:
Web - Config management
Target version:
Severity:
UX impact:
User visibility:
Effort required:
Priority:
Name check:
Fix check:
Regression:

Description

Sometimes, agents get stuck because they all try to lock /var/rudder/cfengine-community/state/cf_lock.tcdb.lock, resulting in a deadlock

In #3928, we implemented a workaround based on the number of agent running, to kill them all if necessary, but here we'll check the age of
/var/rudder/cfengine_community/last_successful_inputs_update

If it is older than 10 minutes, then the agent is stuck, and we kill the .lock file

It happens mainly with cfengine 3.5.x, so targeting to Rudder 2.8


Related issues 4 (0 open4 closed)

Related to Rudder - Bug #3928: Sometimes CFEngine get stuck because of locks on TokyoCabinetReleasedJonathan CLARKE2013-09-13Actions
Related to Rudder - Bug #4548: Correct file opening/closing on cf-agent (backport of 3.5.3 code)RejectedJonathan CLARKEActions
Related to Rudder - Bug #4752: cf_lock.tcdb is not cleaned by check-rudder-agent script when update file is older than 10 minutesReleasedJonathan CLARKE2014-04-11Actions
Related to Rudder - Bug #4769: rudder-agent may be stucked by tokyo cabinet database bloatingReleasedJonathan CLARKE2014-04-23Actions
Actions

Also available in: Atom PDF