custodian (0.9-6) stable; urgency=low

  * Initial implementation of time-period exclusions.

 -- Steve Kemp <steve@bytemark.co.uk>  Tue, 19 Feb 2013 16:44:16 +0000

custodian (0.9-5) stable; urgency=low

  * new release, updated LDAP probe to connect/bind/search.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 13 Feb 2013 10:44:01 +0000

custodian (0.9-4) stable; urgency=low

  * Added duration logging.

 -- Steve Kemp <steve@bytemark.co.uk>  Thu, 23 Jan 2013 11:59:11 +0000

custodian (0.9-3) stable; urgency=low

  * Added redis-protocol tester.

 -- Steve Kemp <steve@bytemark.co.uk>  Mon, 31 Dec 2012 09:55:09 +0000

custodian (0.9-2) stable; urgency=low

  * Include the missing URI library.

 -- Steve Kemp <steve@bytemark.co.uk>  Mon, 24 Dec 2012 11:30:11 +0000

custodian (0.9-1) stable; urgency=low

  * Updated to ensure that protocol tests and the URL-scheme match
    for the case of http & https tests.  This is less surprising.

 -- Steve Kemp <steve@bytemark.co.uk>  Mon, 24 Dec 2012 10:30:01 +0000

custodian (0.8-8) stable; urgency=low

  * Ensure that our notification periods are always integers.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 19 Dec 2012 12:44:21 +0000

custodian (0.8-7) stable; urgency=low

  * New release to ensure that content-matches don't always fail.

 -- Steve Kemp <steve@bytemark.co.uk>  Mon, 17 Dec 2012 11:33:11 +0000

custodian (0.8-6) stable; urgency=low

  * The suppression period, for the mauve alerter, will now vary
    depending upon the time of the day.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 12 Dec 2012 14:16:18 +0000

custodian (0.8-5) stable; urgency=low

  * When sending alerts via Mauve the ID of the alert must be unique,
    per-source.  We thought we'd achieved this by using the input line
    from the parser, but sadly there is a 255 character limit.
    Use the SHA1 hash of the input to satisfy both constraints.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 12 Dec 2012 10:11:01 +0000

custodian (0.8-4) stable; urgency=low

  * The error message is more explicit on DNS failures.

 -- Steve Kemp <steve@bytemark.co.uk>  Tue, 11 Dec 2012 11:00:11 +0000

custodian (0.8-3) stable; urgency=low

  * Mave-alert IDs will be the line from the parser.
  * Updated to use stricter validation on the DNS tests, unknown record
    types will throw an exception.
  * DNS-tests must match results exactly.

 -- Steve Kemp <steve@bytemark.co.uk>  Mon, 10 Dec 2012 13:22:31 +0000

custodian (0.8-2) stable; urgency=low

  * Updated the reporting of http/https failures to show the
    correct protocol-testing type.

 -- Steve Kemp <steve@bytemark.co.uk>  Thu, 6 Dec 2012 11:44:11 +0000

custodian (0.8-1) stable; urgency=low

  * Produce a new package custodian-bytemark
    - This contains a queue monitor.
    - This contains a parser check.
    - Both these scripts are deployed to /usr/bin, and invoked via /etc/cron.d.
  * Don't overwrite the reason for failure when clearing alerts via mauve.

 -- Steve Kemp <steve@bytemark.co.uk>  Thu, 6 Dec 2012 10:22:01 +0000

custodian (0.7-1) stable; urgency=low

  * Set a supression time of 7 minutes on alerts via the mauve
    alerter.

 -- Steve Kemp <steve@bytemark.co.uk>  Thu, 27 Nov 2012 16:45:23 +0000

custodian (0.6-2) stable; urgency=low

  * Set an expiry time on the keys we add to redis.

 -- Steve Kemp <steve@bytemark.co.uk>  Tue, 27 Nov 2012 19:33:19 +0000

custodian (0.6-1) stable; urgency=low

  * Added telnet protocol-test; some of our switches don't
    support SSH.

 -- Steve Kemp <steve@bytemark.co.uk>  Tue, 27 Nov 2012 16:40:50 +0000

custodian (0.5-1) stable; urgency=low

  * Each alerter has a specific destination now.
  * The DNS resolving library is included appropriately.

 -- Steve Kemp <steve@bytemark.co.uk>  Mon, 26 Nov 2012 16:00:00 +0000

custodian (0.4-1) stable; urgency=low

  * Handle the loading of optional libraries more gracefully
    in the alerters.
  * Flush the queue post-installation, or on upgrade.

 -- Steve Kemp <steve@bytemark.co.uk>  Mon, 26 Nov 2012 15:00:51 +0000

custodian (0.3-2) stable; urgency=low

  * Rejigged the HTTP/HTTPS test to avoid SSL validation.
  * Show malformed lines when failing to parse HTTP hosts

 -- Steve Kemp <steve@bytemark.co.uk>  Mon, 26 Nov 2012 12:40:21 +0000

custodian (0.3-1) stable; urgency=low

  * Added a new alerter-type, "redis", which will not raise any
    alerts itself, just store the state of all tests in a redis
    queue.  This can be used to build a status-panel, or similar.

 -- Steve Kemp <steve@bytemark.co.uk>  Mon, 26 Nov 2012 11:22:11 +0000

custodian (0.2-2) stable; urgency=low

  * Explicitly setup a "tube", which is a named queue.
    This means we'll never stomp over the default queue

 -- Steve Kemp <steve@bytemark.co.uk>  Sun, 25 Nov 2012 23:44:32 +0000

custodian (0.2-1) stable; urgency=low

  * Updated to use the user-specified text on test-failure.

 -- Steve Kemp <steve@bytemark.co.uk>  Sun, 25 Nov 2012 22:25:22 +0000

custodian (0.2) stable; urgency=low

  * Removed epoch from all previous versions.
  * Added homepage to the control file.
  * New release.

 -- Steve Kemp <steve@bytemark.co.uk>  Sun, 25 Nov 2012 21:00:21 +0000

custodian (0.1124-1) stable; urgency=low

  * Overhauled much of the code:
    - Test cases are now instantiated via a class-factory pattern.
    - JSON is not used to serialize tests in the queue.
    - The parser and the queue insertion are now decoupled.
    - The test-suite is updated.
    - Many classes were moved into the custodian name-space, and renamed.
    - Many of the protocol tests were re-written to derive from the tcp-test.
    - We use the curb rubygem for all http/https protocol tests.

 -- Steve Kemp <steve@bytemark.co.uk>  Sat, 24 Nov 2012 16:00:16 +0000

custodian (0.1121-1) stable; urgency=low

  * The parsing of the configuration file(s) is now decoupled from
    queue-insertion.
  * The HTTP-status check defaults to 200, unless overridden.

 -- Steve Kemp <steve@bytemark.co.uk>  Tue, 21 Nov 2012 14:55:41 +0000

custodian (0.1120-3) stable; urgency=low

  * Refactor alert code.
  * Ensure that alert-subjects are solely one of : a) hostname b) IP address
    c) a URL.

 -- Steve Kemp <steve@bytemark.co.uk>  Tue, 20 Nov 2012 15:40:30 +0000

custodian (0.1120-2) stable; urgency=low

  * Increased timeout period and retry period
    - To avoid flaps until we rework the alerting with mauve.
  * Fixed most scripts to be ready for ruby 1.9.

 -- Steve Kemp <steve@bytemark.co.uk>  Tue, 20 Nov 2012 14:00:41 +0000

custodian (0.1120-1) stable; urgency=low

  * Added dnsutil.rb - Updated multiping to use it, along with the alerter.
  * Added ruby1.8 as a build-dep to hopefully fix the man-page generation.

 -- Steve Kemp <steve@bytemark.co.uk>  Tue, 20 Nov 2012 10:24:26 +0000

custodian (0.1119-3) stable; urgency=low

  * Many service probes read a banner from a remote server, and then test it
    is sane.  For example looking for "ssh", or "smtp" in the string we read.
    Ensure we read a string prior to testing for a given response.
  * Avoid potential shell-injection attacks against HTTP probes.

 -- Steve Kemp <steve@bytemark.co.uk>  Mon, 19 Nov 2012 16:25:34 +0000

custodian (0.1119-2) stable; urgency=low

  * Fixed the ping-test security hole.

 -- Steve Kemp <steve@bytemark.co.uk>  Mon, 19 Nov 2012 14:15:16 +0000

custodian (0.1119-1) stable; urgency=low

  * Added --error flag to custodian-dequeue.
  * Added SECURITY & README to the docs.

 -- Steve Kemp <steve@bytemark.co.uk>  Mon, 19 Nov 2012 00:11:00 +0000

custodian (0.1118-1) stable; urgency=low

  * The Sundey-before-release Release.

 -- Steve Kemp <steve@bytemark.co.uk>  Sun, 18 Nov 2012 16:19:21 +0000

custodian (0.1117-2) stable; urgency=low

  * Added 'tcp' test with the optional banner.

 -- Steve Kemp <steve@bytemark.co.uk>  Sat, 17 Nov 2012 16:16:16 +0000

custodian (0.1117-1) stable; urgency=low

  * Better alert text.

 -- Steve Kemp <steve@bytemark.co.uk>  Fri, 16 Nov 2012 16:00:16 +0000

custodian (0.1116-1) stable; urgency=low

  * Generate manpages via "make docs", and include those man-pages
    in the generated Debian package.

 -- Steve Kemp <steve@bytemark.co.uk>  Fri, 16 Nov 2012 14:00:41 +0000

custodian (0.1115-4) stable; urgency=low

  * Timeout DNS lookups for alerting purposes.

 -- Steve Kemp <steve@bytemark.co.uk>  Thu, 15 Nov 2012 14:00:41 +0000

custodian (0.1115-3) stable; urgency=low

  * Change the timeouts and repeat counts slightly.
    - Rather than 5 x 10 seconds use 3 x 12 seconds.

 -- Steve Kemp <steve@bytemark.co.uk>  Thu, 15 Nov 2012 10:48:58 +0000

custodian (0.1115-2) stable; urgency=low

  * A HTTP-request might result in a legitimately empty body.

 -- Steve Kemp <steve@bytemark.co.uk>  Thu, 15 Nov 2012 10:28:01 +0000

custodian (0.1115-2) stable; urgency=low

  * Throw exceptions on bogus configuration file entries.
  * Throw exceptions if macros are redefinied.

 -- Steve Kemp <steve@bytemark.co.uk>  Thu, 15 Nov 2012 08:28:08 +0000

custodian (0.1115-1) stable; urgency=low

  * A new day, a new release:
    - Use curl for the HTTP fetching.
    - Remove duplicate HTTPS protocol tester.
    - The details field of all tests will show if a test is in our network.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 14 Nov 2012 22:33:22 +0000

custodian (0.1114-11) stable; urgency=low

  * Show if alerts are inside our network

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 14 Nov 2012 19:30:32 +0000

custodian (0.1114-10) stable; urgency=low

  * New release to pickup the timeout changes.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 14 Nov 2012 16:10:12 +0000

custodian (0.1114-9) stable; urgency=low

  * Added a "custodian-queue --monitor" option, to alert if our queue doesn't
    empty.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 14 Nov 2012 15:05:00 +0000

custodian (0.1114-8) stable; urgency=low

  * Macro definition work when they contain numbers.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 14 Nov 2012 13:45:00 +0000

custodian (0.1114-7) stable; urgency=low

  * Macro definition works for single hosts too.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 14 Nov 2012 11:06:00 +0000

custodian (0.1114-6) stable; urgency=low

  * HTTP/HTTPS tests should work with a missing path.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 14 Nov 2012 10:54:00 +0000

custodian (0.1114-5) stable; urgency=low

  * Removed the -low suffix from alerts.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 14 Nov 2012 10:36:00 +0000

custodian (0.1114-4) stable; urgency=low

  * Don't replace alerts.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 14 Nov 2012 10:12:00 +0000

custodian (0.1114-3) stable; urgency=low

  * Strings are not my friend - interpolate the ID correctly

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 14 Nov 2012 10:00:00 +0000

custodian (0.1114-2) stable; urgency=low

  * Use target_host, not test_host for alerting purposes.
  * Use the correct ID for generating alerts.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 14 Nov 2012 09:51:55 +0000

custodian (0.1114-1) stable; urgency=low

  * Updated alerting via mauve.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 14 Nov 2012 09:13:55 +0000

custodian (0.1114) stable; urgency=low

  * Promoted to live usage:
    - All scripts have a "--help"/"--manual" flag.
    - The alerting has been un-hobbled.

 -- Steve Kemp <steve@bytemark.co.uk>  Wed, 14 Nov 2012 09:13:55 +0000

custodian (0.1113) stable; urgency=low

  * Initial release

 -- Steve Kemp <steve@bytemark.co.uk>  Tue, 13 Nov 2012 15:54:55 +0000