Bug #105006 | provide per-user login session variable settings | ||
---|---|---|---|
Submitted: | 22 Sep 2021 11:46 | Modified: | 4 Jan 13:06 |
Reporter: | Simon Mudd (OCA) | Email Updates: | |
Status: | Verified | Impact on me: | |
Category: | MySQL Server: Options | Severity: | S4 (Feature request) |
Version: | 8.0.26 | OS: | Any |
Assigned to: | CPU Architecture: | Any | |
Tags: | custom login session settings, flexibility, User_attributes |
[22 Sep 2021 11:46]
Simon Mudd
[22 Sep 2021 16:50]
Simon Mudd
No, I don't see this as being the same, the implementation described in the worklog feels different to me. The intent of this FR is that: * settings can be configured per user, perhaps in The User_attributes column, additional JSON fields * as part of the login process (thus a one off action for the session) if configured these user session settings would override the settings populated from the global variables. * after this changes would happen at the session level as before This allows the administrator to optionally custom-configure the session settings which is currently not possible. This avoids the user having to have special connect scripts to setup their environment but still allows this to happen thus overriding the administrator configured settings. We can go into a huge amount of extra complexity about whether we allow some settings to be user-modified or not but that's out of scope for the feature request I'm making. The intent is a SIMPLE way to achieve this, with what should be minimal code-side changes, thus allowing it to be implemented quickly. It would make management of larger environments easier and more flexible.
[22 Sep 2021 16:55]
Simon Mudd
oh and to emphasise a point, https://dev.mysql.com/worklog/task/?id=683 suggests things can be configured in a my.cnf file on the server. That will not work via replication so does not scale if you have a large cluster of servers. I do not want to have to maintain the configuration of X individual servers with Y users. This can be done as I suggest by replication from a single server and distributed via replication to all servers in the cluster / chain pretty much instantaneously so would be a far superior solution I think.
[22 Sep 2021 18:43]
MySQL Verification Team
I indeed read that WL683 too fast.. Let that be irrelevant for this FR. Thanks for the clarification.
[1 Oct 2021 14:26]
Simon Mudd
Another potential use-case would being able to set a new parameter I have requested in bug#105089: range_optimizer_fail_on_memory_exceeded = 1. I could set this on for users where falling back to a table scan would be considered unacceptable but leave it off for more batch oriented users.
[13 Oct 2021 11:09]
Simon Mudd
Also could be used for "max_execution_time".
[4 Jan 11:47]
Simon Mudd
Related: https://bugs.mysql.com/bug.php?id=42035
[4 Jan 13:06]
Simon Mudd
time_zone would also be a good value to use and be able to configure here.