Fluent Bit has an Engine that helps to coordinate the data ingestion from input plugins and calls the Scheduler to decide when it is time to flush the data through one or multiple output plugins. The Scheduler flushes new data at a fixed time of seconds and the Scheduler retries when asked.
Once an output plugin gets called to flush some data, after processing that data it can notify the Engine three possible return statuses:
OK
Retry
Error
If the return status was OK, it means it was successfully able to process and flush the data. If it returned an Error status, it means that an unrecoverable error happened and the engine should not try to flush that data again. If a Retry was requested, the Engine will ask the Scheduler to retry to flush that data, the Scheduler will decide how many seconds to wait before that happens.
The Scheduler provides two configuration options called scheduler.cap and scheduler.base which can be set in the Service section.
scheduler.cap
Set a maximum retry time in seconds. The property is supported from v1.8.7.
2000
scheduler.base
Set a base of exponential backoff. The property is supported from v1.8.7.
5
These two configuration options determine the waiting time before a retry will happen.
Fluent Bit uses an exponential backoff and jitter algorithm to determine the waiting time before a retry.
The waiting time is a random number between a configurable upper and lower bound.
For the Nth retry, the lower bound of the random number will be:
base
The upper bound will be:
min(base * (Nth power of 2), cap)
Given an example where base
is set to 3 and cap
is set to 30.
1st retry: The lower bound will be 3, the upper bound will be 3 * 2 = 6. So the waiting time will be a random number between (3, 6).
2nd retry: the lower bound will be 3, the upper bound will be 3 * (2 * 2) = 12. So the waiting time will be a random number between (3, 12).
3rd retry: the lower bound will be 3, the upper bound will be 3 * (2 * 2 * 2) = 24. So the waiting time will be a random number between (3, 24).
4th retry: the lower bound will be 3, since 3 * (2 * 2 * 2 * 2) = 48 > 30, the upper bound will be 30. So the waiting time will be a random number between (3, 30).
Basically, the scheduler.base determines the lower bound of time between each retry and the scheduler.cap determines the upper bound.
For a detailed explanation of the exponential backoff and jitter algorithm, please check this blog.
The following example configures the scheduler.base as 3 seconds and scheduler.cap as 30 seconds.
The waiting time will be:
1
(3, 6)
2
(3, 12)
3
(3, 24)
4
(3, 30)
The Scheduler provides a simple configuration option called Retry_Limit, which can be set independently on each output section. This option allows us to disable retries or impose a limit to try N times and then discard the data after reaching that limit:
Retry_Limit
N
Integer value to set the maximum number of retries allowed. N must be >= 1 (default: 1)
Retry_Limit
no_limits
or False
When Retry_Limit is set to no_limits
orFalse
, means that there is not limit for the number of retries that the Scheduler can do.
Retry_Limit
no_retries
When Retry_Limit is set to no_retries, means that retries are disabled and Scheduler would not try to send data to the destination if it failed the first time.
The following example configures two outputs where the HTTP plugin has an unlimited number of while the Elasticsearch plugin have a limit of 5 retries: