Github user maoling commented on the issue:
https://github.com/apache/zookeeper/pull/567
@suyogmapara
- **1**.suspicious of the behavious of the method `getCurrentLogSize`
when I test this patch:
- **1.1** set txnLogSizeLimitInKb > preAllocSize
e.g `"-Dzookeeper.txnLogSizeLimitInKb=25600"
"-Dzookeeper.preAllocSize=20480"`
the log block will rise from `20971536` byte to `41943056` byte.it
has changed the logic about preAllocSize and roll over?
- **1.2** set txnLogSizeLimitInKb < preAllocSize
e.g `"-Dzookeeper.txnLogSizeLimitInKb=10240"
"-Dzookeeper.preAllocSize=20480"`
it will increase the number of log blocks insanely
- **2**.still not very clear about what the parameter:
`txnLogSizeLimitInKb` wants to controll and the advantage.
Could you plz explain the following for me? when this happens?
> This is because leader has to scan through the appropriate log file
on disk to find the transaction to start sync from.
---
