IF YOU WOULD LIKE TO GET AN ACCOUNT, please write an
email to Administrator. User accounts are meant only to access repo
and report issues and/or generate pull requests.
This is a purpose-specific Git hosting for
BaseALT
projects. Thank you for your understanding!
Только зарегистрированные пользователи имеют доступ к сервису!
Для получения аккаунта, обратитесь к администратору.
* Fixed log manager so repeated logs do not get logged by default (avoid duplicate log entries)
* Extracted "notifyRestraing" to a method, so it's easier to find/modify it :-)
* Continuing with "from __future__"...
* Added stamp to uniques ids, so we can purge them by date in some cases (Contention)
* DB Migrations to meta groups mixed with stamp of unique id
* Added more logs to some componets
* Added de posibility to authCallback to logout user with redirect
This gives the posibility to to an user service to "reserve" (for now, indefinitely, but will add an "stamp" field to uniqueid database so we can clean up them from time to time) a mac or an name, so other deployed services do not uses them anymore.
One example of this is when a VM gets in error, and machine gets renamed. The mac is them reserved so no more machines clashes with this one reserved.
* Added a couple of needed methods to Storage (so it can pickle and store/retrieve in one step)
* Hanged cleaner implemented, seems to work now
* Restrained service is logged on deployed service as WARN not as arror
* Fixed service view so it accepts meta groups
* Fixed a bug where all HiddenFields where not serializable. Now we can mark a hidden field as serializable (on UserInterface)
* Added push_front to StateQueue
* Changed UniqueIDGenerator so first returns lower values
* Added Service as a log source
* Added state_date when services are marked for removal
* Code cleanup
* Added Cache purge method for owners
* Added improved LDAP Regex Auth
* Added to only process USABLE machines at worker for never getted machines
* Added check for port on server for tunnelers
To do this, the os Manager must inform if it can "process" unused machines (assigned but not in use). If the os manager wants this, there is a background worker,
that works every 10 minutes by default,that will pass every service in that state since more than the last execution of the background worker.
With this, for example, linux and windows os managers, will set this if we mark the action on logout as "remove", and every machine assigned and not in use, for a os manager
marked as "remove", will be removed automatically.
Has to test this extensively!!! :-)
- Disable "global login" page, where all authenticators appears in the list
- Associate an authenticator with an smallName, that can be even the name of the host. Using this, we can access to different authenticators groups using:
* Different host names (that matches against the smallNames and filters them)
* Using the URL http:.../login/[smallName], same as before
* Added "force" parameters to getBool and getInt in Config class
* Fixed xmlrpc view so it works with Django 1.5
* Refactorized Configurations load
* Added "small_name" to authentications, so we can access an specifig auth directly by using url
* Upgraded templates to correctly work with 1.5
* Added scheduled task for checking assigned & never used services, so we can remove them
* Added scheduled task for statistics tables housekeeping (cleans old data & optimizes table if this can be done)
* Added the a friendly description at Job, (for autoregistering)
* Fixed a couple GlobalConfig var name that also has a typo
* Updated migration to include a new index on state_date for user services (new isRestrained method on DeployedServices lookups by this field)
* Added a fix for "broken" Deployed Services that could prevent broker from updating cache of other deployed services if one gets "inoperable". Now we have added a "Restrain" policy, where if a service has 3 or more deploy errors in less than ten minutes, it will be ignored for cache updating until this policy is not satisfied.
