[MDEV-22872] Issues unpacking backup with mbstream Created: 2020-06-11 Updated: 2021-01-20 Resolved: 2021-01-20 |
|
| Status: | Closed |
| Project: | MariaDB Server |
| Component/s: | Backup |
| Affects Version/s: | 10.3.17 |
| Fix Version/s: | 10.3.17 |
| Type: | Bug | Priority: | Major |
| Reporter: | Matej Žerovnik | Assignee: | Robert Bindar |
| Resolution: | Not a Bug | Votes: | 0 |
| Labels: | None | ||
| Environment: |
CentOS 8.x |
||
| Description |
|
When extracting mbstream file, I get an error about missing folder where the original backup was located.
If I manually create folder path, the content is extracted successfully and xtrabackup_info file is placed in the original backup folder.
|
| Comments |
| Comment by Matej Žerovnik [ 2020-06-11 ] |
|
@robertbindar, @Robert Bindar |
| Comment by Robert Bindar [ 2020-06-11 ] |
|
Ack matejzero, thanks again! I will investigate and let you know. |
| Comment by Tim Yan [ 2020-10-12 ] |
|
It seems the mbstream is reusing the --extra-lsndir value used in the original backup command line. If the location was a tmp directory and deleted, the mbstream will fail. It may also fail if you unpack the same backup file twice, because the xtrabackup_info will be there already. The only info we found related to the tmp directory is in the xtrabackup_info itself and its copies. |
| Comment by Matej Žerovnik [ 2021-01-04 ] |
|
Is there any news regarding a fix? This is hurting us more and more and a quick feedback would be useful, so we know if we should wait or search for a different solution. Thanks. |
| Comment by Robert Bindar [ 2021-01-20 ] |
|
Hei Tim and Matej, very sorry for being unresponsive here, for some reason Jira didn't notify me for your messages. I have a quick instinct to say I merged myself a Pull Request fixing this issue a while ago, but it's probably in a newer MariaDB version. I'll investigate this tomorrow first thing in the morning. |
| Comment by Matej Žerovnik [ 2021-01-20 ] |
|
I had a look on GitHub and noticed that there was indeed a commit into a newer version in 10.3 train. We upgraded to newer version and it works. I don't know which version fixed it as I can't find the commit now. |
| Comment by Robert Bindar [ 2021-01-20 ] |
|
Hey Matej, what a nice thing that the bug was solved for you with just upgrading. The contribution I was speaking about is: https://github.com/MariaDB/server/commit/c0db3fe6dabd2cb0df9bf532f54f9e8d233fa499 |
| Comment by Robert Bindar [ 2021-01-20 ] |
|
Closing this for now as it seems the issue was resolved. Feel free to reopen if the bug can be reproduced again. Thanks again for keeping an eye on this. |