Graphite

配置

由于 SQLite3 的局限性，在事件比较多时会很慢甚至报错。推荐使用 PostgreSQL。

Web

cd /opt/graphite/conf
cp graphite.wsgi.example graphite.wsgi
# edit graphite.wsgi

cd /opt/graphite/webapp/graphite
cp local_settings.py.example local_settings.py
# edit local_settings.py
# 注意修改时区！它并不会默认使用系统时区

PYTHONPATH=/opt/graphite/webapp django-admin.py syncdb --settings=graphite.settings

# 创建 events 需要
setfacl -m u:graphite:rwx /opt/graphite/storage

nginx 配置

server {
    listen       80;
    server_name  xxx;

    location / {
            root /opt/graphite/webapp;
            try_files $uri @uwsgi;
    }

    location /static {
            alias /opt/graphite/webapp/content;
    }

    location @uwsgi {
            uwsgi_pass unix:/var/run/graphite/graphite.sock;
            include uwsgi_params;
    }
}

uwsgi 配置

[uwsgi]
master = true
workers = 10
socket = /var/run/graphite/graphite.sock
chdir = /opt/graphite/conf
virtualenv = /opt/graphite
wsgi-file = graphite.wsgi
chmod-socket = 660

# uid = graphite
# gid = graphite
# plugins = python2

mkdir -p /var/run/graphite
chown graphite: /var/run/graphite /opt/graphite/storage /opt/graphite/storage/graphite.db /opt/graphite/storage/log/webapp
usermod -aG graphite http

uwsgi --ini /etc/uwsgi/graphite.ini --daemonize2 /var/log/graphite.log

carbon

配置：

cd /opt/graphite/conf
cp carbon.conf.example carbon.conf
# edit carbon.conf
cp storage-schemas.conf.example storage-schemas.conf
# edit storage-schemas.conf
cp storage-aggregation.conf.example storage-aggregation.conf
# edit storage-aggregation.conf

chown carbon: /opt/graphite/storage/log
chown carbon: /opt/graphite/storage
chown carbon: /opt/graphite/storage/whisper

启动：

sudo -u carbon /opt/graphite/bin/carbon-cache.py start

自启动

uwsgi --ini /etc/uwsgi/graphite.ini --daemonize2 /var/log/graphite.log
su carbon -c "/opt/graphite/bin/carbon-cache.py start"

statsd

可在 storage-schemas.conf 中加入如下设置：

[statsd]
pattern = ^stats
retentions = 10s:2d,1m:14d,10m:365d,1h:1800d

在 storage-aggregation.conf 中如下设置：

[min]
pattern = \.(min|lower)$
xFilesFactor = 0.1
aggregationMethod = min

[max]
pattern = \.(max|upper)$
xFilesFactor = 0.1
aggregationMethod = max

[sum]
pattern = \.count$
xFilesFactor = 0
aggregationMethod = sum

[default_average]
pattern = .*
xFilesFactor = 0.5
aggregationMethod = average

注意

whisper 存储数据时，数据的时间戳会根据 retention 取到不大于其的时间点上。比如提交的数据时间戳是 1464767098，最小 retention 是 10s，那么 whipser 会将数据存储到 1464767090。

启动 carbon 时如遇到以下错误：

twisted.python.usage.UsageError: The specified reactor cannot be used, failed with error: reactor already installed.
See the list of available reactors with --help-reactors

可注释掉 /opt/graphite/lib/carbon/util.py:61

        twistd_options.append("--reactor=epoll")

参见

• statsd
• Tools That Work With Graphite — Graphite 0.10.0 documentation
• InfluxDB: 另一存储后端
• Grafana: 另一展示界面

文档

• Configuring Carbon — Graphite documentation
• using graphite events – devops type thoughts

外部链接

• obfuscurity. - Migrating Graphite from SQLite to PostgreSQL