[checkmk-commits] Check_MK Git: check_mk: FIX Improved performance of WATO slave push with a large user base

Module: check_mk Branch: master Commit: 52c448c91a6d5d743026f7ed38e425ef6d33334b URL: http://git.mathias-kettner.de/git/?p=check_mk.git;a=commit;h=52c448c91a6d5d… Author: Lars Michelsen <lm(a)mathias-kettner.de> Date: Mon Dec 16 12:09:06 2013 +0100 FIX Improved performance of WATO slave push with a large user base When having a large number of users (several thousands) and pushing them to a slave site, the user database hook was executed twice, which resulted in a senseless duplicate calculation of things. This has been resolved now. Conflicts: .bugs/1018 ChangeLog --- .bugs/980 | 7 +++++-- .werks/330 | 12 ++++++++++++ ChangeLog | 1 + web/htdocs/userdb.py | 1 + web/plugins/wato/userdb.py | 5 ----- 5 files changed, 19 insertions(+), 7 deletions(-) diff --git a/.bugs/980 b/.bugs/980 index 303d566..d285c64 100644 --- a/.bugs/980 +++ b/.bugs/980 @@ -1,9 +1,9 @@ Title: Hooks users-saved called duplicate when doing WATO sync Component: wato -State: open +Class: nastiness +State: done Date: 2013-10-08 18:24:47 Targetversion: 1.2.3i3 -Class: nastiness When doing a replicated WATO sync, then the hook users-saved is being called twice. We should elimiate that. Especially because @@ -28,3 +28,6 @@ there might be involved lengthy LDAP syncs [Tue Oct 08 09:23:31 2013] [warn] [client 127.0.0.1] CALLING HOOK activate-changes [Tue Oct 08 09:23:32 2013] [warn] [client 127.0.0.1] FINISHED HOOK activate-changes [Tue Oct 08 09:23:32 2013] [warn] [client 127.0.0.1] RESTARTED + +2013-12-16 12:06:40: changed state open -> done +Removed the snapshot-pushed hook. diff --git a/.werks/330 b/.werks/330 new file mode 100644 index 0000000..47eb7ae --- /dev/null +++ b/.werks/330 @@ -0,0 +1,12 @@ +Title: Improved performance of WATO slave push with a large user base +Level: 1 +Component: wato +Class: fix +State: unknown +Version: 1.2.5i1 +Date: 1387192025 +Targetversion: future + +When having a large number of users (several thousands) and pushing them to a +slave site, the user database hook was executed twice, which resulted in a +senseless duplicate calculation of things. This has been resolved now. diff --git a/ChangeLog b/ChangeLog index b5d9350..5bacde5 100644 --- a/ChangeLog +++ b/ChangeLog @@ -42,6 +42,7 @@ * 0242 FIX: Parameters for clustered services can now be configured on the cluster host... * 0309 FIX: Trying to prevent read/write conflicts with a large user base... * 0311 FIX: Fixed "Inventory failed" message when trying an inventory on clusters via WATO... + * 0330 FIX: Improved performance of WATO slave push with a large user base... Notifications: * 0108 FIX: Prevent service notification on host alerts... diff --git a/web/htdocs/userdb.py b/web/htdocs/userdb.py index 55905d1..6e23468 100644 --- a/web/htdocs/userdb.py +++ b/web/htdocs/userdb.py @@ -664,6 +664,7 @@ def hook_login(username, password): # Is called on: # a) before rendering the user management page in WATO # b) a user is created during login (only for this user) +# c) Before activating the changes in WATO def hook_sync(connector_id = None, add_to_changelog = False, only_username = None, raise_exc = False): if connector_id: connectors = [ get_connector(connector_id) ] diff --git a/web/plugins/wato/userdb.py b/web/plugins/wato/userdb.py index 0d5663e..43608fd 100644 --- a/web/plugins/wato/userdb.py +++ b/web/plugins/wato/userdb.py @@ -30,8 +30,3 @@ def sync_pre_activate_changes(_unused): if 'wato_pre_activate_changes' in config.userdb_automatic_sync: userdb.hook_sync() api.register_hook('pre-activate-changes', sync_pre_activate_changes) - -def sync_snapshot_pushed(): - if 'wato_snapshot_pushed' in config.userdb_automatic_sync: - userdb.hook_sync() -api.register_hook('snapshot-pushed', sync_snapshot_pushed)