Cannot balance data after a balance attempt failed on an offline storage host

[wey-gu]

Describe the bug (must be provided)

A clear and concise description of what the bug is.

Your Environments (must be provided)

2.5.0

How To Reproduce(must be provided)

Steps to reproduce the behavior:

1. offline a storage host

in this step, the host is still presented in show hosts

2. start a data balance
3. data balance failed
4. storage host was not shown in show host

Now, host is not shown in show hosts

5. balance data

Then it's observed the old failed job id was returned

Expected behavior

It should start a new balance data job.

Additional context

https://nebulagraph.slack.com/archives/CJB79PQG2/p1632827375002600

[learn2Pro]

any solution? i still can not balance data

[critical27]

I can't open the slack... There are a few questions:

1. After the first step offline a storage host, the offline hosts is on/off in show hosts?
2. The third step: why we failed? We can see the reason in metad's log.
3. The forth step: storage host was not shown in show host would take more than one week's time, sure about that?

[critical27]

PS: there is a command called balance data reset plan, which will remove the last failed balance plan. After that command succeed, new balance could be triggered.

[wey-gu]

I can't open the slack... There are a few questions:

1. After the first step offline a storage host, the offline hosts is on/off in show hosts?
2. The third step: why we failed? We can see the reason in metad's log.
3. The forth step: storage host was not shown in show host would take more than one week's time, sure about that?

I updated the original description to add detailed info for step 1 and 4.
Please @learn2Pro add more information :)

PS: there is a command called balance data reset plan, which will remove the last failed balance plan. After that command succeed, new balance could be triggered.

As I recall, @learn2Pro only BALANCE DATA STOP was tried, I think BALANCE DATA RESET PLAN is the command that's needed here, can you try it? I think this issue could be closed as BALANCE DATA RESET PLAN exists.

I will create a docs PR to include BALANCE DATA RESET PLAN in https://docs.nebula-graph.com.cn/2.5.1/3.ngql-guide/18.operation-and-maintenance-statements/2.balance-syntax/ rather than just in https://docs.nebula-graph.com.cn/2.5.1/8.service-tuning/load-balance/#_4

[learn2Pro]

solved