Permalink
Switch branches/tags
Nothing to show
Commits on Oct 15, 2013
  1. Updated version for 4.0.1 release

    Eric Stanley committed Oct 15, 2013
  2. Updated Changelog for 4.0.1 release

    Eric Stanley committed Oct 15, 2013
  3. Reverted commit f99a9a7 which set check_interval to 1 if it was confi…

    Eric Stanley committed Oct 15, 2013
    …gured as zero.
    
    Setting check_interval to 0 is useful when you don't want to perform
    scheduled active checks, but want to allow on demand checks.
  4. Added compiler flags in RPM spec file to reduce compiler noise (Dan W…

    Eric Stanley committed Oct 15, 2013
    …ittenberg)
    
    Here's another update to the nagios.spec that puts back something I
    removed due to complaints it broke el5 builds and didn't have time
    before 4.0 to fix the right way. It basically allows el6 based
    systems (with newer gcc) to suppress some of the dumb warnings
    during compile time, so easier to spot real issues.
Commits on Oct 11, 2013
  1. Removed unused variable (Dan Wittenberg)

    Eric Stanley committed Oct 11, 2013
    ...and got rid of compiler warning at the same time.
Commits on Oct 5, 2013
  1. logging: Corrected order of arguments when logging unknown hosts/serv…

    Eric Stanley committed Oct 5, 2013
    …ices
Commits on Oct 4, 2013
  1. base/nebmod: Added logging of failure in dlclose() call

    Eric Stanley committed Oct 4, 2013
    We recently had a problem with a buggy glibc implementation of dlclose
    causing module unloading to misbehave. Chances are having been
    informed of the result of the dlclose() call would've been helpful in
    debugging that.
    
    Signed-off-by: Anton Lofgren
  2. downtime: Downtime initialized before retention data read

    Eric Stanley committed Oct 4, 2013
    The downtime data needs to be initialized before the retention data is
    read so when there was a downtime in effect when core was stopped it will
    be recreated on restart.
  3. test/test-downtime: Fixes to single-run code

    Eric Stanley committed Oct 4, 2013
    Previously the code run when the --single-run option would only ever run a
    flexible check. Now it will run the check requested and also display the
    check requested.
  4. test/test-downtime: Updates for core 4 and fixes

    Eric Stanley committed Oct 4, 2013
    The test-downtime.pl script has been updated to check the version of
    Nagios Core it is running. If the major version is 4 or greater, all
    calculations that previously involved reaper times now ignore the
    reaper times. In addition, problems were resolved that were related
    to the length of downtimes.
Commits on Sep 27, 2013
  1. Patches to make RPM build again (Dan Wittenberg)

    Eric Stanley committed Sep 27, 2013
Commits on Sep 24, 2013
  1. Make cgi's build again

    Eric Stanley committed Sep 24, 2013
    The recent changes to the downtime core functions broke the build
    for cgi's. This restores them to good working order.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
Commits on Sep 20, 2013
  1. Updated Changelog for core 4 release

    Eric Stanley committed Sep 20, 2013
  2. Merge branch 'master' of ssh://git.code.sf.net/p/nagios/nagioscore

    Eric Stanley committed Sep 20, 2013
    * 'master' of ssh://git.code.sf.net/p/nagios/nagioscore:
      downtime: Ensure that scheduled_downtime_depth never drops below zero
      downtime: Handle error checking in downtime_add()
      downtime: Handle scheduled_downtime_list from downtime_{add,remove}
      downtime: Make the downtime list a doubly-linked one
      downtime: Refactor delete_downtime()
      downtime: Use downtime_compar() when insert-sorting downtime
      downtime: Simplify the insertion sort somewhat
      downtime: Make downtime lookups a 16384 times more efficient
      core: Replace half of xdata/xdddefault.c with a 5-line helper
      common/downtime: Simplify a few functions
  3. downtime: Ensure that scheduled_downtime_depth never drops below zero

    ageric committed with Eric Stanley Sep 20, 2013
    If it does, subsequent downtimes may cause the depth to be bumped up
    to zero, which will cause the notification machinery to not consider
    the host or service to be in downtime at all, which in turn means
    that notifications will be sent out while downtime is actually in
    effect. It's unclear how one triggers this without modifying the
    retention data file, but it has happened a few times in the wild,
    and this rather simple patch efficiently protects against it.
    
    Signed-off-by: Anton Lofgren <alofgren@op5.com>
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  4. downtime: Handle error checking in downtime_add()

    ageric committed with Eric Stanley Sep 20, 2013
    We're supposed to have all the information we need there, so we might
    as well do all error checking there and get rid of the validation
    routine in xdata/xdddefault.c, which is then reduced to setting
    next_downtime_id = 1 in case nothing else sets it when we initialize.
    
    Moving that small task to common/downtime.c as well means we can get
    rid of the entire xdata/xdddefault.c, which is a very nice bonus.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  5. downtime: Handle scheduled_downtime_list from downtime_{add,remove}

    ageric committed with Eric Stanley Sep 19, 2013
    It makes a lot more sense to handle all storage from the API-ish
    functions that we can test separately, so refactor stuff so they
    end up there so we can test the storage system.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  6. downtime: Make the downtime list a doubly-linked one

    ageric committed with Eric Stanley Sep 19, 2013
    This makes the algorithms for insertion and removal a lot more
    pleasant, as we can abuse find_downtime() without having to bother
    about locating the previous entry for efficient removals.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  7. downtime: Refactor delete_downtime()

    ageric committed with Eric Stanley Sep 19, 2013
    The only error condition we can run into is when we don't find the
    downtime we're supposed to delete, so return ERROR immediately if
    that happens and avoid some unnecessary indentation and a (very)
    long if() clause.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  8. downtime: Use downtime_compar() when insert-sorting downtime

    ageric committed with Eric Stanley Sep 19, 2013
    It's important that we maintain the sort order when we're inserting
    downtime, or we'll end up with inconsistent states if we sometimes
    sort using qsort() and downtime_compar(), and other times by using
    a different comparison function.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  9. downtime: Simplify the insertion sort somewhat

    ageric committed with Eric Stanley Sep 19, 2013
    If scheduled_downtime_list is NULL, there's no need for us to enter
    the iteration loop, so we can handle it the same we would as if we
    were deferring downtime sorting and just prepend the downtime to
    the (empty) scheduled_downtime_list.
    
    We can do the same trick if the downtime_compar() function tells us
    that the new downtime sorts before the first entry in the list.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  10. downtime: Make downtime lookups a 16384 times more efficient

    ageric committed with Eric Stanley Sep 19, 2013
    For a given value of "efficient" at least. Since we're using a fanout
    table, we'll be using some memory in exchange for the fewer operations,
    but at least lookup operations have decent practical performance now.
    
    While at it, we sneak in the embryo of an API, which we can build on
    later when refactoring the code into something sensible.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  11. core: Replace half of xdata/xdddefault.c with a 5-line helper

    ageric committed with Eric Stanley Sep 19, 2013
    It's mostly useless, and the half-dozen or so other downtime adding
    functions can surely pick up the slack, should there be any.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  12. common/downtime: Simplify a few functions

    ageric committed with Eric Stanley Sep 18, 2013
    In preparation for a large-scale refactoring, this is the kind of
    boilerplate crap I rather not be distracted by.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  13. Prepare for 4.0.0 release

    Eric Stanley committed Sep 20, 2013
  14. downtime: Ensure that scheduled_downtime_depth never drops below zero

    ageric committed Sep 20, 2013
    If it does, subsequent downtimes may cause the depth to be bumped up
    to zero, which will cause the notification machinery to not consider
    the host or service to be in downtime at all, which in turn means
    that notifications will be sent out while downtime is actually in
    effect. It's unclear how one triggers this without modifying the
    retention data file, but it has happened a few times in the wild,
    and this rather simple patch efficiently protects against it.
    
    Signed-off-by: Anton Lofgren <alofgren@op5.com>
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  15. sample-config: Update with new options for Nagios 4

    Daniel Wittenberg committed with ageric Sep 17, 2013
    Most people tend to look at the sample configuration when setting up
    their nagios configurations. This patch makes it passably simple for
    them to find and get a brief description of all the various options,
    although some of them are experimental and shouldn't really be used.
    
    Message-by: Andreas Ericsson <ae@op5.se>
    Signed-off-by: Daniel.Wittenberg <Daniel.Wittenberg@ipsoft.com>
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  16. downtime: Handle error checking in downtime_add()

    ageric committed Sep 20, 2013
    We're supposed to have all the information we need there, so we might
    as well do all error checking there and get rid of the validation
    routine in xdata/xdddefault.c, which is then reduced to setting
    next_downtime_id = 1 in case nothing else sets it when we initialize.
    
    Moving that small task to common/downtime.c as well means we can get
    rid of the entire xdata/xdddefault.c, which is a very nice bonus.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  17. downtime: Handle scheduled_downtime_list from downtime_{add,remove}

    ageric committed Sep 19, 2013
    It makes a lot more sense to handle all storage from the API-ish
    functions that we can test separately, so refactor stuff so they
    end up there so we can test the storage system.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  18. downtime: Make the downtime list a doubly-linked one

    ageric committed Sep 19, 2013
    This makes the algorithms for insertion and removal a lot more
    pleasant, as we can abuse find_downtime() without having to bother
    about locating the previous entry for efficient removals.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  19. downtime: Refactor delete_downtime()

    ageric committed Sep 19, 2013
    The only error condition we can run into is when we don't find the
    downtime we're supposed to delete, so return ERROR immediately if
    that happens and avoid some unnecessary indentation and a (very)
    long if() clause.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  20. downtime: Use downtime_compar() when insert-sorting downtime

    ageric committed Sep 19, 2013
    It's important that we maintain the sort order when we're inserting
    downtime, or we'll end up with inconsistent states if we sometimes
    sort using qsort() and downtime_compar(), and other times by using
    a different comparison function.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  21. downtime: Simplify the insertion sort somewhat

    ageric committed Sep 19, 2013
    If scheduled_downtime_list is NULL, there's no need for us to enter
    the iteration loop, so we can handle it the same we would as if we
    were deferring downtime sorting and just prepend the downtime to
    the (empty) scheduled_downtime_list.
    
    We can do the same trick if the downtime_compar() function tells us
    that the new downtime sorts before the first entry in the list.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>
  22. downtime: Make downtime lookups a 16384 times more efficient

    ageric committed Sep 19, 2013
    For a given value of "efficient" at least. Since we're using a fanout
    table, we'll be using some memory in exchange for the fewer operations,
    but at least lookup operations have decent practical performance now.
    
    While at it, we sneak in the embryo of an API, which we can build on
    later when refactoring the code into something sensible.
    
    Signed-off-by: Andreas Ericsson <ae@op5.se>