[Bug 1709670] Re: logrotate never recovers if the statefile is corrupted

[Taylor Lewick]

We tested this patched package version on ubuntu 14.04, via instructions
found https://wiki.ubuntu.com/Testing/EnableProposed.

We tested by manually creating a 2.0 GB file via fallocate under the
nginx log directory, and instructing the logrotate nginx config file to
rotate any file > 1.0 GB.

This worked, and then I manually corrupted the logrotate status file by 
1) removing the top line, creating a new test log file, and verifying logrotate ran.
2) adding a line break in a file other than the top line, creating a new test log file, and verifying logrotate ran.

Patch Behavior appears to be working as designed.

Thank you,
Taylor

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to logrotate in Ubuntu.
https://bugs.launchpad.net/bugs/1709670

Title:
  logrotate never recovers if the statefile is corrupted

Status in logrotate package in Ubuntu:
  Fix Released
Status in logrotate source package in Trusty:
  Fix Committed
Status in logrotate source package in Xenial:
  Fix Committed
Status in logrotate source package in Zesty:
  Fix Committed
Status in logrotate source package in Artful:
  Fix Released
Status in logrotate package in Debian:
  New

Bug description:
  [Impact]

  logrotate never recovers if the statefile is corrupted unless you
  remove it or fix the corruption by hand.

  Impact scenarios :

  - System could eventually run out of disk space on a separate
  partition if mounted in "/var" or specifically "/var/log" or even
  worst if "/var/log" is on the same partition as "/" it could create
  even more damage if by any chance the partition is running out of free
  space.

  - System keep updating the same files over and over, creating large
  size logfiles.

  - ...

  [Test Case]

  - Install logrotate
  - Run "/etc/cron.daily/logrotate" ## The first logrotate run will generate the statefile "var/lib/logrotate/status"
  - Modify "/var/lib/logrotate/status" by removing the first line in order to corrupt the file
  - Re-run "/etc/cron.daily/logrotate" and one will get the following error : "error: bad top line in state file /var/lib/logrotate/status" every time you run logrotate

  Unless you remove the statefile and start again or fix the corruption
  by hand.

   * Additionally, I will run the /path_to_source/test/test script as a
  dogfooding that does ~72 tests.

  [Regression Potential]

   * Risk of potential regression is low, and IMHO couldn't be worst
  than the actual situation where logrotate simply doesn't recover from
  a corrupt statefile.

   * The current patch does recover (after verification) and has been
  through some upstream CI validation, community feedbacks, et al.

   * Additionally, I will run the /path_to_source/test/test script as a
  dogfooding that does ~72 tests.

  [Other Info]

  * Upstream commit:
  https://github.com/logrotate/logrotate/commit/b9d82003002c98370e4131a7e43c76afcd23306a

  * Upstream bug:
  https://github.com/logrotate/logrotate/issues/45

  * Debian bug:
  https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=871592

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/logrotate/+bug/1709670/+subscriptions