Essential TDH Community Edition Ops Guide: Common Issues & Solutions

Common Operational Issues

Component configuration files are located at /etc/<component>/conf and component JAR packages at /usr/lib/<component>.

Log Locations

Component logs can be found under /var/log/<component_instance>, where <component_instance> is the instance name.

Adjust Manager Log Size and Count

Edit /etc/transwarp-manager/master/log4j.properties and modify the *.MaxFileSize (log file size) and *.MaxBackupIndex (number of retained logs) parameters.

Disk Space Warning

When the Manager UI shows a low‑disk‑space alert, check root usage with du -h --max-depth=1|grep G and clean large directories such as /var/log, /hadoop/data, /ngmr, or move files to other disks.

Start/Stop Services via kubectl (Example: Inceptor)

# Check Inceptor service status
kubectl get po -owide | grep inceptor
# Stop Inceptor Metastore role
kubectl delete -f /var/lib/transwarp-manager/master/content/resources/services/inceptor1/inceptor-metastore.yaml
# Start Inceptor Metastore role
kubectl create -f /var/lib/transwarp-manager/master/content/resources/services/inceptor1/inceptor-metastore.yaml
# Restart Metastore role
kubectl delete pod <inceptor_metastore_pod>

Notes: start/stop must be executed on the Manager node; restart can be run on any node using the -s flag to specify the TOS Server IP.

Health‑Check Time Offset Alert

The alert indicates unsynchronized node clocks; synchronize the time across nodes.

Manually Sync Node Time

systemctl stop ntpd
ntpdate <NTP Server hostname>
systemctl start ntpd

Disk Write Failure

# Check disk space
df -h
# If full, migrate data
# Check disk health
smartctl -H <target_disk>
# Check read speed
hdparm -Tt <target_disk>
# Check inode usage
df -i

If inode usage is saturated, delete unnecessary temporary files to free inodes.

Connection Refused in Logs

Verify network connectivity between nodes, ensure /etc/hosts entries are correct, and check that the firewall is disabled. Use systemctl status firewalld, then systemctl stop firewalld and systemctl disable firewalld if needed.

Component Link Unreachable

Make sure the URL uses the node’s hostname and that the hostname‑IP mapping is present in the local hosts file (Windows: C:\Windows\system32\drivers\etc\hosts; Linux: /etc/hosts).

Manager Cannot Connect to Agent (Ask Timeout)

# Test connectivity to agent hostname
nc -v <agent_hostname> 10208
# Test connectivity to agent IP
nc -v <agent_ip> 10208
# Ping the agent
ping <agent_ip>
# Check firewall status
systemctl status firewalld
# Stop and disable firewall if active
systemctl stop firewalld
systemctl disable firewalld

If the firewall is off but the timeout persists, increase remote-start-timeout in /etc/transwarp-manager/master/application.conf to 60 seconds and restart the Manager.

Can Hostname Be Modified After Adding a Node?

No. Hadoop identifies nodes by hostname/FQDN; changing it is equivalent to removing the node and adding a new one.

License Service Issues

In the Manager UI, go to Management → License to view service status; if the service is down (e.g., more than two License Server nodes are offline), click “Start”. The license expiration date is also shown on this page.

Summary

This article covers common TDH Community Edition operational scenarios, including configuration file locations, log management, disk‑space alerts, service start/stop via kubectl, time synchronization, disk health checks, network and firewall troubleshooting, hostname resolution, and license service handling.