We used nginx proxy to cache lots of static files. Our monitoring tool
showed that the write I/O on the server spike once a day on 4am GMT time
and server will slow down significantly because of the high IO. I
couldn’t
find any cron that can cause this. I am guessing it might be the nginx
cache manager was clearing the disk cache once a day.
We have cache setting like: proxy_cache_path /cache levels=2:2:2
keys_zone=cache:2000m inactive=1d max_size=400000m;
On Thu, Aug 09, 2012 at 07:02:13PM -0700, Don Zhou wrote:
We used nginx proxy to cache lots of static files. Our monitoring tool
showed that the write I/O on the server spike once a day on 4am GMT time
and server will slow down significantly because of the high IO. I couldn’t
find any cron that can cause this. I am guessing it might be the nginx
cache manager was clearing the disk cache once a day.
No, it’s not nginx cache manager. It removes files as soon as
they become inactive and/or max_size reached.
We have cache setting like: proxy_cache_path /cache levels=2:2:2
keys_zone=cache:2000m inactive=1d max_size=400000m;
Any idea? Thanks!
Something like “top -mio” / iotop might be helpfull. Time
suggests most likely it’s some daily periodic task.
You right! I found it! It is the mlocate.cron come with centos install.
It
basic run updatedb on the file system and set the nice to be +19. I have
so
many files in my caching directory and updatedb will consume almost all
the
IO. I disabled my caching directory in updatedb.conf.
Thanks for the help.
This forum is not affiliated to the Ruby language, Ruby on Rails framework, nor any Ruby applications discussed here.