CentOS7 SELinux阻止MongoDB启动的问题

问题描述：

在新装的CentOS7上，安装了MongoDB3.4，挂载了一个大的数据盘后，修改/etc/mongo.conf，将配置文件中的log和data目录放在新的数据盘下，并修改文件的访问权限。改完后的mongo.conf:

# mongod.conf

# for documentation of all options, see:
#   http://docs.mongodb.org/manual/reference/configuration-options/

# where to write logging data.
systemLog:
  destination: file
  logAppend: true
  path: /data/mongodb/log/mongod.log

# Where and how to store data.
storage:
  dbPath: /data/mongodb/data
.....

文件权限：

# ls -alh
drwxr-xr-x.  5 mongod mongod 4.0K 11月  1 14:53 mongodb

# cd mongodb
# ls -alh
drwxr-xr-x. 3 mongod mongod 4.0K 11月  9 19:08 data
drwxr-xr-x. 2 mongod mongod 4.0K 11月  9 19:06 log
drwxr-xr-x. 2 mongod mongod 4.0K 11月  1 14:54 run

执行systemctl start mongod命令后，查看状态发现并没有启动，查看/var/log/message，发现以下错误

Nov  9 06:06:44 [localhost] setroubleshoot: failed to retrieve rpm info for /data/mongodb/run/mongod.pid
Nov  9 06:06:44 [localhost] setroubleshoot: SELinux is preventing /usr/bin/mongod from write access on the file /data/mongodb/run/mongod.pid. For complete SELinux messages run: sealert -l f7148e11-b126-401e-ba9f-a9a87c1e54ae
Nov  9 06:06:44 [localhost] python: SELinux is preventing /usr/bin/mongod from write access on the file /data/mongodb/run/mongod.pid.#012#012*****  Plugin restorecon (94.8 confidence) suggests   ************************#012#012If you want to fix the label. #012/data/mongodb/run/mongod.pid default label should be default_t.#012Then you can run restorecon.#012Do#012# /sbin/restorecon -v /data/mongodb/run/mongod.pid#012#012*****  Plugin catchall_labels (5.21 confidence) suggests   *******************#012#012If you want to allow mongod to have write access on the mongod.pid file#012Then you need to change the label on /data/mongodb/run/mongod.pid#012Do#012# semanage fcontext -a -t FILE_TYPE ‘/data/mongodb/run/mongod.pid‘#012where FILE_TYPE is one of the following: afs_cache_t, initrc_tmp_t, mongod_log_t, mongod_tmp_t, mongod_var_lib_t, mongod_var_run_t, puppet_tmp_t, user_cron_spool_t.#012Then execute:#012restorecon -v ‘/data/mongodb/run/mongod.pid‘#012#012#012*****  Plugin catchall (1.44 confidence) suggests   **************************#012#012If you believe that mongod should be allowed write access on the mongod.pid file by default.#012Then you should report this as a bug.#012You can generate a local policy module to allow this access.#012Do#012allow this access for now by executing:#012# ausearch -c ‘mongod‘ --raw | audit2allow -M my-mongod#012# semodule -i my-mongod.pp#012

从提示中可以看出是SELinux的防护功能，阻止了访问。

解决过程：

/var/log/message中的信息看起来比较费劲，里面有一句提示：

For complete SELinux messages run: sealert -l e73ba9e8-f74d-4835-9b53-85667546b28c

根据提示执行：

# sealert -l e73ba9e8-f74d-4835-9b53-85667546b28c
SELinux is preventing /usr/bin/mongod from write access on the directory /data/mongodb/log.

*****  Plugin catchall_labels (83.8 confidence) suggests   *******************

If you want to allow mongod to have write access on the log directory
Then 必须更改 /data/mongodb/log 中的标签
Do
# semanage fcontext -a -t FILE_TYPE ‘/data/mongodb/log‘
其中 FILE_TYPE 为以下内容之一：mongod_log_t, mongod_tmp_t, mongod_var_lib_t, mongod_var_run_t, tmp_t, var_lib_t, var_log_t, var_run_t。
然后执行：
restorecon -v ‘/data/mongodb/log‘

*****  Plugin catchall (17.1 confidence) suggests   **************************
......

上面提示输出中已经包含了，解决方法：

# semanage fcontext -a -t mongo_log_t ‘/data/mongodb/log‘
# restorecon -v ‘/data/mongodb/log‘
restorecon reset /data/mongodb/log context unconfined_u:object_r:unlabeled_t:s0->unconfined_u:object_r:mongod_log_t:s0

上面命令执行完毕后，就解决了/data/mongodb/log目录的文件权限问题。

同样的方法，再解决/data/mongodb/data和/data/mongodb/run目录的问题。

启动mongod，问题解决。

Nov  9 06:08:51 [localhost] systemd: Starting High-performance, schema-free document-oriented database...
Nov  9 06:08:51 [localhost] systemd: Started High-performance, schema-free document-oriented database.
Nov  9 06:08:51 [localhost] mongod: about to fork child process, waiting until server is ready for connections.
Nov  9 06:08:51 [localhost] mongod: forked process: 18218
Nov  9 06:08:51 [localhost] mongod: child process started successfully, parent exiting

P.S. 除了上面通过提示信息解决问题外，还有一个比较暴力的方法，直接关闭SELinux，但是不太建议。

# setenforce 0
# getenforce
Permissive

上面是临时关闭，如果是永久关闭，就需要编辑/etc/selinux/config文件，将SELINUX=enforcing改为SELINUX=disabled，但是只有重启后才会发挥作用。