Upgrading TiDB from 7.5.0 to 7.5.1: Dashboard Login No Longer Required

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

Original topic: tidb 7.5.0升级到7.5.1 dashborad登录不用了

| username: zhanggame1

[TiDB Usage Environment] Production Environment / Testing / Poc
[TiDB Version] 7.5.1
[Reproduction Path] What operations were performed when the issue occurred
[Encountered Issue: Problem Phenomenon and Impact]
Error reported when logging in as root after upgrade

Tried restarting PD and then logging in again

| username: zhanggame1 | Original post link

Restarting the cluster is fine, but why does the dashboard have issues with rolling updates?

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

Are you using HAProxy for proxying? Did you enable transparent mode? When the error occurred previously, did you enable the restriction that PD nodes must log in through VIP?

| username: lemonade010 | Original post link

If any of the following situations occur, login may fail:

  • TiDB root user does not exist
  • PD is not started or cannot be accessed
  • TiDB is not started or cannot be accessed
  • Incorrect root password
| username: Kongdom | Original post link

Looking at the release documentation, this is the only relevant one.

| username: zhanggame1 | Original post link

Direct connection, no proxy, both DB upgrades have this issue.

| username: zhanggame1 | Original post link

It’s possible that the parameters have changed.

| username: FutureDB | Original post link

Rolling upgrades can sometimes still cause issues. When we previously upgraded from V6.1.0 to V6.5.4, TiDB Dashboard had no problems, but we encountered a situation where we couldn’t log into Grafana using the original username and password. We had to resolve it by resetting the user password.

| username: TIDB-Learner | Original post link

If the database is accessible, try changing the password and make it more complex.

| username: Jellybean | Original post link

When upgrading, did you confirm that the PD nodes were fully restarted?

You can check the pd.log to investigate the logs before and after the restart; there should be some clues.

| username: zhanggame1 | Original post link

After the upgrade, it didn’t work, so I manually reloaded PD. It still didn’t work, but finally, after restarting the cluster, it was fine.

| username: Jellybean | Original post link

The restart operation involves rotating stop and start.
The reload operation, as I recall, involves stopping, then refreshing the configuration files and binary files, and then starting again.

If the upgrade has already been completed, there is no essential difference between the two. You can check the logs for more details.

I speculate that some cluster metadata information is loading slowly and is not ready yet, causing it to be inaccessible. After some time, it completes.

| username: TiDB_C罗 | Original post link

Can I log in using MySQL?

| username: zhanggame1 | Original post link

Restart means shutting everything down and then starting it again, which is different from reload.

| username: 小龙虾爱大龙虾 | Original post link

:joy_cat: Check if there is any additional information in the logs, and test if the MySQL client can log in.