Deployment Issues: How to Deploy Single IP with Multiple Ports for SSH

Note:
This topic has been translated from a Chinese forum by GPT and might contain errors.

Original topic: 部署问题,单IP多端口ssh如何部署

| username: bryanz

[TiDB Usage Environment] Production
[TiDB Version] v7.1.1
[Reproduction Path] What operations were performed when the problem occurred
[Encountered Problem: Deployment Issue, How to Deploy Single IP Multiple Port SSH]
[Resource Configuration] kvm nat mode, the host is accessed via kvm IP + port, how to deploy the tidb cluster.
[Attachment: Screenshot/Log/Monitoring]
Host 192.168.1.3 192.168.1.4 192.168.1.5
Each host has 3 virtual machines, no IP in the 192.168.1.0 subnet. The virtual machines are in nat mode and need to be accessed through host port mapping.

| username: zhanggame1 | Original post link

I didn’t quite understand, could you describe it more? How many virtual machines are there? Why is there a multi-port SSH issue with the deployment?

| username: MrSylar | Original post link

KVM can set up an internal virtual network bridge, and use that IP to deploy the cluster.

| username: MrSylar | Original post link

It was possible to do this on Huawei Cloud before.

| username: bryanz | Original post link

Host machines: 192.168.1.3, 192.168.1.4, 192.168.1.5
Each host machine has 3 virtual machines, none of which have IPs in the 1 subnet. The virtual machines are in NAT mode and need to be accessed through port mapping on the host machine.

| username: bryanz | Original post link

Boss, the online environment is already fixed. KVM is using NAT mode, and the host machine’s network segment IP, which is the public IP, is only a few.

| username: zhanggame1 | Original post link

Just choose a virtual machine as a jump server, mainly depending on what problem you need to solve.

| username: redgame | Original post link

If it works, it can be done; if it doesn’t, it can’t be done.

| username: bryanz | Original post link

It’s connected and installed, but it won’t start.

alertmanager-server:13001 alertmanager alertmanager-server 13001/13002 linux/x86_64 Down /zol/tidb-data/alertmanager-9093 /zol/tidb-deploy/alertmanager-9093
grafana-serve:13001 grafana grafana-serve 13001 linux/x86_64 Down - /zol/tidb-deploy/grafana-3000
pd-01:10001 pd pd-01 10001/10002 linux/x86_64 Up|L /zol/tidb-data/pd-2379 /zol/tidb-deploy/pd-2379
pd-02:10001 pd pd-02 10001/10002 linux/x86_64 Down|UI /zol/tidb-data/pd-2379 /zol/tidb-deploy/pd-2379
pd-03:10001 pd pd-03 10001/10002 linux/x86_64 Down /zol/tidb-data/pd-2379 /zol/tidb-deploy/pd-2379
monitoring-server:13001 prometheus monitoring-server 13001/12020 linux/x86_64 Down /zol/tidb-data/prometheus-8249 /zol/tidb-deploy/prometheus-8249
tidb-01:11001 tidb tidb-01 11001/11002 linux/x86_64 Down - /zol/tidb-deploy/tidb-4000
tidb-02:11001 tidb tidb-02 11001/11002 linux/x86_64 Down - /zol/tidb-deploy/tidb-4000
tidb-03:11001 tidb tidb-03 11001/11002 linux/x86_64 Down - /zol/tidb-deploy/tidb-4000
tikv-01:12001 tikv tikv-01 12001/12002 linux/x86_64 Up /zol/tidb-data/tikv-20160 /zol/tidb-deploy/tikv-20160
tikv-02:12001 tikv tikv-02 12001/12002 linux/x86_64 Up /zol/tidb-data/tikv-20160 /zol/tidb-deploy/tikv-20160
tikv-03:12001 tikv tikv-03 12001/12002 linux/x86_64 Up /zol/tidb-data/tikv-20160 /zol/tidb-deploy/tikv-20160

| username: bryanz | Original post link

The IPs of the three host machines are 192.168.1.2, 192.168.1.3, and 192.168.1.4.
The IPs of the virtual machines on the host machines are 192.168.100.2, 192.168.100.3, 192.168.100.4, and 192.168.100.5.
The IPs of the virtual machines on the host machines are 192.168.100.2, 192.168.100.3, 192.168.100.4, and 192.168.100.5.
The IPs of the virtual machines on the host machines are 192.168.100.2, 192.168.100.3, 192.168.100.4, and 192.168.100.5.
The nodes are deployed on virtual machines, and each virtual machine is connected via the host machine’s IP and mapped port.
Can this mode be deployed?

| username: bryanz | Original post link

The image you provided is not visible. Please provide the text you need translated.

| username: bryanz | Original post link

The image is not available for translation. Please provide the text content directly.

| username: 像风一样的男子 | Original post link

Before deployment, check for potential risks in the cluster:
tiup cluster check ./topology.yaml --user root [-p] [-i /home/root/.ssh/gcp_rsa]

| username: bryanz | Original post link

Checked and found no issues.

| username: 像风一样的男子 | Original post link

Check the network between each node to see if the firewall is not turned off.

| username: bryanz | Original post link

The network is normal. It just won’t start.

| username: zhanggame1 | Original post link

It’s best not to use hostnames for deployment, as there are quite a few issues. Deploying on virtual machines is definitely fine; I always use virtual machines for installation, and communication is also handled through the virtual switch within the virtual machines.

| username: bryanz | Original post link

There’s no other way. The virtual machines on the host in the online environment all have the same network segment and some have the same IP, so we can only deploy using IP plus port and host method.

| username: tidb狂热爱好者 | Original post link

First understand virtual machines before working on TiDB, and use an internal network IP address.

| username: bryanz | Original post link

Using internal IP addresses with tiup will result in duplicate IPs being assigned.