Extremely high CPU usage about gin HOT 14 CLOSED

Thanks @mikegleasonjr!

from gin.

Glad my bump resulted in so much awesome 👍 🍻 😄

from gin.

How much file do you have in your path (recursively)?
It is possible that delay of 500ms (between two checks for changes) could be too small for you (if fs/cpu is too busy and can't process app requests in between).
We should consider adding that as a cli parameter (or consider using file/dirwatchers for detecting changes) if that proves to be your problem.

@codegangsta do you want me to do that?

from gin.

I have this problem as well, caused by one of those bloody node_modules directories behemoths where I run Gin. Makes my CPU go nuts after a while.

The ability to specify which directories should be watched would be a huge improvement!

from gin.

+1 for that. node_modules is a real pain

from gin.

+1 mine too '___')

$ find . | wc
  25589   27322 1452620
$ find . -type 'd' | wc
   2052    2173   84120
$ find . -iname '*.go' | wc
    144     144    3617
$ find .git | wc
     83      83    2374
$ find resources/ | wc # this folder will never ever be compiled
  24311   26044 1361828
$ find resources/ -type 'd' | wc
   1935    2056   78388

an exclusion flag will be nice, for example:

gin -i -b /tmp/test --exclude resources/

from gin.

👍

Is this project dead? Any alternatives?

from gin.

#83 fixes this issue. Sadly it its unmerged.

I have a fork with that PR merged to master over at https://github.com/vektah/gin so it can be go get github.com/vektah/gin

from gin.

Bump...

from gin.

I ended up using fswatch:

Not a shell guru so bare with me:

$ fswatch -0 -or --include ".*\.go$" --exclude ".*" . | xargs -0 -t -n 1 -I {} go install -v github.com/your/server &
$ fswatch -0 -or $GOPATH/bin/server | xargs -0 -t -n 1 -I {} sh -c 'killall server; $GOPATH/bin/server &> optional.log &' &

The first line rebuilds and installs your package when *.go changes, the second listens for a change of your installed server. (sometimes rebuilding does not yield a new binary, just a small optimization)

When I'm done, I killall fswatch and killall server. I was using prm anyway so I had startup and stop scripts at my disposition to automate that.

Lighter, more portable, faster, easier to run, etc..

from gin.

#83 is merged!

from gin.

I have excluded all huge directories with the excludeDir but still cpu is about 40% on idle .(
mac os 10.12.4

from gin.

oops there was just a typo. All fixed.

from gin.

--excludeDir unnecessaryDir especially directories with node_modules helps a lot (14% cpu usage to 0-2%)

from gin.