You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Mikhail Pochatkin (Jira)" <ji...@apache.org> on 2023/04/12 12:20:00 UTC

[jira] [Updated] (IGNITE-18581) Bootstrap Configuration: don't save bootstrap configuration in vault

     [ https://issues.apache.org/jira/browse/IGNITE-18581?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Mikhail Pochatkin updated IGNITE-18581:
---------------------------------------
    Epic Link: IGNITE-19280

> Bootstrap Configuration: don't save bootstrap configuration in vault
> --------------------------------------------------------------------
>
>                 Key: IGNITE-18581
>                 URL: https://issues.apache.org/jira/browse/IGNITE-18581
>             Project: Ignite
>          Issue Type: New Feature
>          Components: build
>            Reporter: Mikhail Pochatkin
>            Assignee: Mikhail Pochatkin
>            Priority: Major
>              Labels: ignite-3
>             Fix For: 3.0.0-beta2
>
>          Time Spent: 2h
>  Remaining Estimate: 0h
>
> *Motivation*
> *There is no single mechanism for changing the configuration*
> As described above currently a user needs to modify two different configuration files (vars.env and ignite-conf.conf) in order to configure a node. But these configuration files perform different tasks. At the moment, the vars.env file specifies the arguments that are necessary to start the node and read the configuration properties, if necessary, from the vault. This is the name of the node, the path to the vault work directory, the path to the configuration file, and the node finder argument. If the first three have a clear purpose, because without them it is impossible to start reading the configuration, then the last argument is quite controversial and its necessity does not seem obvious.
> *Mixing of Ignite node configuration and state*
> Currently after the first node start provided bootstrap configuration is stored in the local node vault. After the node is restarted, the new bootstrap configuration is merged with what was saved to the local vault; if the user deleted or commented out any property in the configuration file before restarting the node, then it will be applied from the saved configuration from the last run. This can cause a lot of confusion on the part of the user because the application applies some of the properties for the user. So, in conclusion, the current mechanism of configuration storage looks more like migrating configuration to node state and we have the stateful node after start.
> This ticket reqiore changes to stop saving the starting configuration to the local vault. It is proposed to make the configuration file on the local fs the only source of truth, and the node should also read this file every time it starts. In such a case, changing the file by the user and restarting the node will apply those changes. These changes should be validated before applying and saving.
> Manual changes in config file should be applied only after node restart. In case when a user modified a configuration file manually (modified config file) with some errors after restart, the node should fail to start with a good explanation of reasons and the user should understand which kind of error presented in the config file.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)