From 076ac1aa14be66f3eb0c8d11c6f912a2839af269 Mon Sep 17 00:00:00 2001 From: Jakob Borg Date: Wed, 11 May 2016 11:44:45 +0000 Subject: [PATCH] users/faq: Mention scanning and syncthing-inotify in relation to CPU usage GitHub-Pull-Request: https://github.com/syncthing/docs/pull/167 --- users/faq.rst | 10 ++++++++++ 1 file changed, 10 insertions(+) diff --git a/users/faq.rst b/users/faq.rst index 1c3a8f175..d6c233815 100644 --- a/users/faq.rst +++ b/users/faq.rst @@ -129,6 +129,11 @@ Why does it use so much CPU? #. There is a certain amount of housekeeping that must be done to track the current and available versions of each file in the index database. +#. By default Syncthing uses periodic scanning every 60 seconds to detect + file changes. This means checking every file's modification time and + comparing it to the database. This can cause spikes of CPU usage for large + folders. + Hashing, compression and encryption cost CPU time. Also, using the GUI causes a certain amount of extra CPU usage to calculate the summary data it presents. Note however that once things are *in sync* CPU usage should be @@ -140,6 +145,11 @@ Syncthing should use at any given moment. For example, ``GOMAXPROCS=2`` on a machine with four cores will limit Syncthing to no more than half the system's CPU power. +To reduce CPU spikes from scanning activity, use a filesystem notifications +plugin. This is delivered by default via Synctrayzor, Syncthing-GTK and on +Android. For other setups, consider using `syncthing-inotify +`_. + Should I keep my device IDs secret? ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~