mykola/retracker
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

add notes

Browse source
This commit is contained in:
mykola2312 2024-11-02 03:36:15 +02:00
parent 7f75a1ac77
commit fa7e340ad7
1 changed files with 11 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

11
notes.txt
View file

@ -42,5 +42,16 @@ Local tracker cache and destination tracker peers:
after absence in N announce responses from destination tracker.
If we don't flush peer list - it will indefinitely grow bigger and bigger.
Server architecture 1:
Tracker class has synchronized methods for access by running threads. Also Tracker has list of running threads, that are either
TrackerServer or TrackerJob. Tracker is responsible for starting and stopping them.
Server architecture 2:
There is some other entity like Application class that handles running threads, making Tracker responsible for only one thing
abd that thing is tracking peers of torrents. This is preferred design choice since we can then export tracking and scraping
functionality in separate interfaces, since UDP announce protocol has scrape method, whilst HTTP doesn't. As for now, current Tracker
class implementation doesn't provide any other information for Torrent besides its InfoHash and PeerSet, which is built from local announces
and announces to destination trackers.
Maven exec with args:
mvn exec:java -Dexec.args=''