Monitoring Flussonic with Retroview¶
You can monitor the status of your Flussonic server(s) and analyze the quality of the incoming stream using the Retroview
How to open Retroview?¶
First, you need to log in to your personal account
Expand the Retroview tab
Here you will see two types of monitoring: Input monitoring and Server Status .
Input monitoring¶
Analyzes the quality of the stream that comes to Flussonic.
Total Input Bitrate
Displays the total incoming bitrate of the stream.
Input Availability
Shows the status of the input stream:
- Offline – No input stream
- Good – Input stream is present and there are no input errors
- Bad – Input stream is present, but input errors are detected
Top N Streams with Input Errors
Displays the top streams by error count – these are streams that require attention.
DVR Recording Issues
Shows how DVR segments are written. Segment quality types:
- Fast – Segments written faster than ½ of segment duration
- Slow – Written slower than ½, but faster than 90% of duration
- Delayed – Slower than 90%, DVR cannot record the stream
- Failed – Write failed due to a storage error
- Skipped – Segment wasn't written due to storage delay; it could have been written if storage was faster, but it was removed from memory
Per-stream details¶
Stream Input Error Details
Breakdown of input errors per stream:
- lost_packets – Lost packets during the period
- broken_payload – Corrupted payloads
- dropped_frames – Dropped frames
- ts_stuck_restarts – TS stuck connection restarts (e.g., in RTSP)
- desync – MPEG-TS synchronization lost
- ts_pat – Missing PAT
- ts_service_lost – TS service lost
- src_404 / src_403 / src_500 – HTTP errors from source
- crashed – Source crashes
- dropped_packets – Packets dropped for various reasons
- no_signal – Frames dropped due to signal loss
- duplicate – Frame duplicated due to input slowness
- ts_duplicate – Repeated frame timestamp
- cpu_stall – Frame dropped due to high CPU load
- adaptation_broken – Packets with adaptation field larger than packet size
- ts_scrambled – Scrambled TS packets
- ts_pmt – PMT not received after 0.5 seconds
- ts_cc – Non-contiguous MPEG-TS continuity counters
- ts_tei – MPEG-TS packets with Transport Error Indicator
- ts_psi_checksum – Broken PSI checksum
- broken_pes_count – PES started not from startcode
- discarded_buffer_count – Discarded large ES buffers
- ts_crashed – Crashes inside MPEG-TS decoder
- too_large_dts_jump – Large DTS jump, flush required
- errors_pid_lost – Lost MPEG-TS PID
- rtp_pt_reject – RTP packets with wrong payload type
- discarded_not_allowed_nal_count – Disallowed NAL units in non-interleaved mode
- agent_conn_failed – The agent was unable to open the requested connection. These errors may indicate problems opening the TCP socket on the agent or remote host is unreachable
- agent_out_of_memory – These errors indicate that agent does not have enough memory to establish a connection to the remote host
- agent_buffer_overrun – These errors indicate that agent does not have enough buffer size to handle outgoing traffic
- agent_invalid_request – These errors indicate that agent is receiving invalid requests
- agent_unknown_errors – Unknown agent errors counter
- resync_count_jump - Number of times the source performed an unexpected timestamp jump, indicating unannounced discontinuities or resets in the stream timing
Stream Input Bitrate
Incoming bitrate per stream.
Stream Input Warning Details
Warnings that are corrected by the server:
- ts_stuck – TS stuck issue restarts
- sr_ts_stuck – SR packets with repeated RTP timestamp
- sender_clock_deviation – Sender clock ahead/behind server time
- ts_goes_backwards – Time jumped back on the channel
- ts_jump_forward – Time jumped forward
- no_marker_mode_flag – Decoder works in no marker mode
- fu_pattern_is_broken_count – Broken FU pattern
- fu_has_both_start_end_bits_count – FU with both Start and End bits set
- fu_end_then_middle_workaround_count – FU workaround applied
- dts_stuck – Repeated DTS
- dts_goes_backwards – DTS jumped back
- dts_jump_forward – DTS jumped forward
DVR Recording Issues (Per Stream)
DVR segment quality for an individual stream.
Stream Input Source
Displays which source is currently used by the stream.
Stream Input Source Switches
Number of times the input source changed for the stream.
Server Status¶
It provides statistics on the status of each server. The following parameters can be seen on the graphs: CPU Load, Disk Utilization %, Disk Write Metrics, and Scheduler Load etc.
Top N servers with performance issues
Highlights servers with performance issues based on CPU, memory, or scheduler load exceeding 85% at any time. Pay attention if any of these thresholds are exceeded, as it indicates potential instability.
System Resources:
CPU Load – Shows the server’s average CPU usage
Scheduler Load – Shows system-level scheduler pressure
Memory % Usage – Percentage of used RAM
DVR Errors – Error when doing operations on DVR segments
Disk I/O:
Disk Utilization % – Percentage of disk usage
Disk Write Metrics – Metrics for disk write operations
Disk Read Metrics – Metrics for disk read operations
Streaming & Clients:
Streams – Number of active streaming sessions
Clients – Number of connected clients
Server Bandwidth:
in_media_bw – Total incoming bandwidth used for media traffic
in_system_bw – Total incoming bandwidth used by the system
out_media_bw – Total outgoing bandwidth used for media transmission
out_system_bw – Total outgoing bandwidth used by the system
GPU Stats:
GPU Utilization – GPU usage percentage
GPU Temperature – Current temperature of the GPU
GPU Encoder Load – GPU load related to video encoding
GPU Decoder Load – GPU load related to video decoding
GPU Memory Usage – Memory usage of the GPU
GPU CUDA Load – GPU load based on CUDA cores
Alerts Dashboard¶
Available alerts to create
Allows you to create alerts based on various metrics and events related to servers and streams. Supports selecting specific servers and streams, as well as configuring a contact point for notifications.
Step-by-step instructions:
1. Select a server – At the top of the dashboard, choose the server where the alert should apply. Some alerts support the All option to apply across all servers.
2. Select streams – Required only for alerts related to specific streams.
3. Choose an alert – In the Available alerts to create section, select the type of alert you want to create.
4. Configure a contact point –
• For email, you can create a contact directly in the Create email contact point section
• For others (Telegram, Slack, etc.), use Grafana’s Alerting → Contact points page
5. Enter an alert name – Provide a unique name for your alert
6. Choose an evaluation interval (pending period) –
This is the time Grafana waits before triggering the alert once the condition is met.
Specify it in formats like:
• 10s (10 seconds)
• 30s (30 seconds)
• 1m (1 minute – recommended default)
• 5m (5 minutes)
• 1h (1 hour)
If you're unsure which value to use, the default and most optimal choice is 1m
7. Confirm – If all parameters are correctly filled in, the alert will be created and added to the active alerts list
Available Alerts:
cpu_load
– Triggers when the average CPU load exceeds 85% over the past hour
Requires selecting a specific server
scheduler_load
– Triggers when the average system scheduler load exceeds 85% over the past hour
Requires selecting a specific server
memory_usage
– Triggers when average memory usage exceeds 85% over the past hour
Requires selecting a specific server
disk_io
– Triggers when disk utilization exceeds 85% over the past hour
Requires selecting a specific server
streams_drop
– Triggers for Watcher when more than 10% of streams are dropped during a given time period
Does not require server selection – applies to all Watcher servers
stream_dead
– Triggers when a stream that previously had input suddenly stops
You can select a specific server or use All to apply globally
selected_stream_dead
– Similar to stream_dead, but only monitors selected streams
You can select a specific server or All, and must select which streams to monitor
flapping_streams
– Triggers when a stream temporarily loses input and recovers more than 3 times within 3 hours
This may indicate network issues, an unstable input source, or server-side instability
You can select a specific server or use All to monitor all streams for flapping behavior
selected_stream_flapping
– Similar to flapping_streams, but only monitors selected streams
You must choose which streams to monitor, and can also select a specific server or use All
Flapping may be caused by network problems, source-side interruptions, or delivery infrastructure issues
input_availability_raise_offline
– Triggers when the number of offline streams increases by the specified percentage, as shown on the input_availability graph.
Can be created for all servers or for a specific one.
After selecting this alert type, an additional field will appear where you must specify the percentage increase in offline streams that should trigger the alert
input_availability_raise_bad
– Triggers when the number of streams with input errors increases by the specified percentage, as shown on the input_availability graph.
Can be created for all servers or for a specific one.
After selecting this alert type, an additional field will appear where you must specify the percentage increase in bad streams that should trigger the alert
Create Email Contact Point
Allows you to create a contact point for sending alert notifications to email addresses. Convenient to use directly from the dashboard without going into Grafana settings.
How to create an Email Contact Point:
1. Contact name – Enter a unique name for the contact point. This name will appear in the recipient selection dropdown.
2. Email addresses – Enter one or more email addresses to receive alert notifications.
To enter multiple addresses, separate them with commas.
Example:
example@mymail.com,example2@mymail.com
After saving, the contact point will appear in the list and be available for selection when creating alerts.
Alerts List
Displays all created alerts. Allows you to view key information at a glance and manage existing alerts.
Displayed information:
Alert name – Unique name assigned during creation
Folder – The folder or group to which the alert belongs
Status – If the alert is paused, a paused label will appear
Managing alerts:
Each alert has a ✕ button. Clicking it will delete the alert.
More details:
For full details and alert logic, go to:
Grafana → Alerting → Alert rules
Contact Points List
Displays a list of all created contact points used for sending alert notifications. Allows you to review and remove any contacts.
Displayed information:
Type – The type of contact point (e.g., Email, Webhook, etc.)
Name – The unique name given during contact creation
Managing contact points:
Each contact point has a ✕ button. Clicking it will delete the contact.
More details:
To view or configure contact points in Grafana, go to:
Grafana → Alerting → Contact points