Flussonic Central 23.11
С выходом версии 23.11 представляем вам ряд классных усовершенствований Flussonic Central. Теперь мы используем технологию кластерного захвата Central при подключении стримеров, расширили возможности создания URL-адресов архивных фрагментов, доработали запросы к узлам видеоаналитики и внедрили механизм защиты от создания потоков, превышающих заданное значение.
И это не все! В обновлении 23.11 вы найдете информацию обо всех изменениях в Central!
Flussonic Central 23.11
- #31311 (new) The peer config on streamers (for the old cluster ingest mechanism) is now dropped when adding a streamer to the cluster managed by Central to take all advantages of the Central’s cluster ingest technology.
- #31438 (new) Added the ability to get stream’s layout history showing the reason for the layout change. This will help you understand why the stream moves from one streamer to another and how often this happens as well as configure your streams and streamers so that to minimize the layout changes.
- #32098 (new) Implemented the new approach to preparing URLs of archive fragments on servers in the cluster allowing the use of Public Payload URL or API URL before the hostname. This will allow to handle the situations when the hostname is not FQDN-compliant.
- #32645 (new) Added the check for video analytics nodes being online or offline before fetching episodes from it to reduce the number of requests by skipping offline nodes.
- #33186 (new) Implemented streams reallocation when a new streamer is added to Central. Reallocation takes place only if it will improve overall system performance compared to the previous state.
- #33407 (new) Added the check for the number of streams with the same streamer hostname as in the created stream to avoid creating more streams than the
channel_limit for that streamer.
- #33469 (new) The streamers statistics is now returned as part of response to GET /config request to allow monitoring of the cluster status and capacity.
- #33470 (new) The agent statistics is now returned as part of response to GET /config request to allow monitoring of the cluster status and capacity.
- #33682 (new) Implemented the use of streaming_endpoint parameter for indication of the server where the video from IP camera is currently ingested to allow the use of Watcher Client UI on VSAAS.IO and other cluster installations.
- #33890 (new) Added the ability to manage the stream and streamer tags. The tags will allow more control over the stream allocation when adding/replacing/upgrading/removing streamers. You will also be able to determine the server priorities for reallocating streams in case of failure or assign part of the streamers in the cluster to a specific set of streams.
- #34012 (new) The uptime, id, and started_at parameters are now returned in response to the ‘GET /config` request for more insight in the streamer status.
- #34018 (new) Added the ability to collect Central metrics via GET /config in OpenMetrics format for use with Prometheus.
- #32494 (fixed) Fixed the issue with double slash in URLs returned by Central’s balancer for redirects.
- #33015 (fixed) Fixed the slow operation of cursor in the person management API.
- #33072 (improved) Improved the protection against memory leaks when connecting Agents to enhance the service stability.
- #33318 (fixed) Fixed HTTP 500 errors and timeouts when collecting events from Central that may have resulted in skipped analytics events.
- #33522 (improved) The Central auth backend for analytics now uses
allow_tokens parameter instead of requesting Central as an application auth backend to reduce load on the Central server by allowing streamers authenticate analytics nodes themselves.
- #33542 (fixed) Fixed the periodical spontaneous restarts of Central with
panic in logs.
- #33558 (fixed) Fixed the issue with online/offline and provisioned/not provisioned streams not returned in stats of the cluster config.
- #33566 (fixed) Fixed the issue with periodical unexpected resets of the channel_limit parameter.
- #33574 (fixed) Fixed the issue with some streams with Agents not being able to run with
keeplive drop message in Agent’s logs because of conflicts at handling messages from Agents in Central.
- #33617 (improved) Disabled validation of streamer hostname to allow domain names ending with a dot (
- #33621 (fixed) Fixed the issue with
episodes_expiration incorrectly returned in response to
- #33769 (fixed) Fixed the excessive PostgreSQL logs interfering with the system operation and preventing new events reception.
- #33843 (fixed) Fixed the issue with streams not played in the player on the Overview tab in the Central Admin UI.
- #33871 (fixed) Fixed the HTTP 404 error returned in response to person management API.
- #33915 (improved) Refactored the Central’s config generation process. The logic stay the same: environment variables take precedence over the
central.conf config file.
- #33960 (improved) The streamer polling interval decreased to 30 seconds by default (compared to once every two seconds) to reduce the load on the Central server; the OBSERVER_POLLING_INTERVAL config parameter added to adjust it when needed.
- #34061 (improved) The listeners config is now returned in response to
GET /config request.
- #34093 (improved) Added the support for specifying API key of analytics nodes in the API_URL parameter as
http://api-key@hostname to allow the Vision servers to operate in a cluster managed by Central.
- #34263 (fixed) Fixed the issue with outdated information about the archive parts location on streamers not deleted and needless inputs provisioned for the streams causing unnecessary load on the system.