mirrors/tendermint
1
0
Fork 0
mirror of https://github.com/tendermint/tendermint.git synced 2026-09-20 15:04:22 +00:00
Code Issues Packages Projects Releases Wiki Activity

tools/tm-bench: Don't count the first block if its empty (#1948)

Browse Source 
* tools/tm-bench: Don't count the first block if its empty

* Try melekes suggestion

* Fix getting the start time so the first block is no longer empty

* Fix changelog entry

* Make code smell better

* tools/tmbench: Fix the end time being used for statistics calculation

Previously we were using the time at which all connections closed in statistics, not
the time after {duration} seconds.

* Use waitgroups for starting up
This commit is contained in:
Dev Ojha
2018-07-13 12:28:19 +04:00
committed by Anton Kaliaev
parent bbf2bd1d81
commit 262c8daa04
6 changed files with 79 additions and 37 deletions
Download Patch File Download Diff File Expand all files Collapse all files
tools/tm-bench/README.md
+17 -6
View File
@@ -51,15 +51,26 @@ with the last command being in a seperate window.
## How stats are collected
These stats are derived by having each connection send transactions at the
specified rate (or as close as it can get) for the specified time. After the
specified time, it iterates over all of the blocks that were created in that
time. The average and stddev per second are computed based off of that, by
specified rate (or as close as it can get) for the specified time.
After the specified time, it iterates over all of the blocks that were created
in that time.
The average and stddev per second are computed based off of that, by
grouping the data by second.
To send transactions at the specified rate in each connection, we loop
through the number of transactions. If its too slow, the loop stops at one second.
If its too fast, we wait until the one second mark ends. The transactions per
second stat is computed based off of what ends up in the block.
through the number of transactions.
If its too slow, the loop stops at one second.
If its too fast, we wait until the one second mark ends.
The transactions per second stat is computed based off of what ends up in the
block.
Note that there will be edge effects on the number of transactions in the first
and last blocks.
This is because transactions may start sending midway through when tendermint
starts building the next block, so it only has half as much time to gather txs
that tm-bench sends.
Similarly the end of the duration will likely end mid-way through tendermint
trying to build the next block.
Each of the connections is handled via two separate goroutines.