If maxFileSize is not specified when using SizeAndTimeBasedFNATP, then some logs are not written. Roughly 1 out of 10 logs are lost in my case, were I generate a lof logs (approx 15 per ms).

Maybe the cause is that the threshold size is parsed as 0, and it's constantly reopening the file?

Example:
<timeBasedFileNamingAndTriggeringPolicy
class="ch.qos.logback.core.rolling.SizeAndTimeBasedFNATP">
</timeBasedFileNamingAndTriggeringPolicy>

thanks