无法通过systemd启动mongos服务，MongoDB分片集群问题求助

Let’s walk through the most common issues that could block your mongos from starting via systemd, based on your cluster setup:

1. Fix YAML Syntax Errors in mongos.conf #

The most likely culprit here is invalid YAML formatting in your configuration file. YAML relies on strict indentation and structure—your current config has a syntax issue where configDB isn’t properly nested under sharding. Here’s the corrected version:

sharding:
  configDB: config0/mongo-config-1:27017,mongo-config-2:27017,mongo-config-3:27017
net:
  bindIp: 0.0.0.0
# Add essential missing fields
pidFilePath: /var/run/mongodb/mongos.pid
logPath: /var/log/mongodb/mongos.log
logAppend: true

Make sure to use spaces (not tabs) for indentation—YAML doesn’t tolerate tabs.

2. Verify Config Replica Set Health & Connectivity #

Mongos can’t start unless it can reach a healthy config replica set. Follow these checks:

• On the mongos node, test DNS resolution for your config nodes:

  ping mongo-config-1
  ping mongo-config-2
  ping mongo-config-3
  

  If pings fail, update /etc/hosts with the IPs of your config nodes or fix your DNS setup.
• Connect to any config node and validate the replica set status:

  mongosh --host mongo-config-1:27017
  

  Run rs.status() in the shell—ensure the config0 replica set is initialized, has a PRIMARY node, and all nodes show a stateStr of PRIMARY or SECONDARY (config replica sets shouldn’t use arbiters, your 3-node setup is correct).

3. Validate Systemd Service Configuration #

Your mongos.service file is missing critical sections. A complete working example should look like this:

[Unit]
Description=MongoDB Shard Router (mongos)
After=network.target

[Service]
User=mongodb
Group=mongodb
ExecStart=/usr/bin/mongos --config /etc/mongos.conf
PIDFile=/var/run/mongodb/mongos.pid
Restart=always
RestartSec=5
LimitNOFILE=64000

[Install]
WantedBy=multi-user.target

Key checks:

• Confirm ExecStart points to the correct path of your mongos binary (use which mongos to verify).
• Ensure PIDFile matches the pidFilePath in mongos.conf.
• Fix permissions for log and PID directories:

  chown -R mongodb:mongodb /var/log/mongodb/ /var/run/mongodb/
  

4. Check Logs for Exact Error Details #

When systemd fails to start a service, logs will tell you exactly why. Run this command to get detailed error output:

journalctl -u mongos.service -xe

You can also check the mongos log file (specified in mongos.conf) for startup errors—this will reveal issues like config parsing failures, connection timeouts to the config set, or permission denials.

5. Test Manual Mongos Startup #

To isolate whether the issue is with systemd or mongos itself, start mongos manually without systemd:

sudo -u mongodb mongos --config /etc/mongos.conf

If this works, the problem lies in your systemd service configuration. If it fails, the error output will directly point to the root cause (e.g., invalid config, unreachable config nodes).

内容的提问来源于stack exchange，提问作者Hemant Mann