I have production server running on 2.3.1 on HA and i wanted to upgrade the version to 3.0.7 with out any downtime. I can create additional server with latest version and restore the backup from 2.3.1 to 3.0.7 but the challenge for me is, i will lose all the data during the migration time frame ( while running the re-deployment of API pointing to new server ). I really appreciate if any one have idea on how i can accomplish this with out having downtime on my production environment. Currently server is configured as HA and its pointed to application via LB
Asked
Active
Viewed 76 times
1 Answers
0
I don't think it's possible. Safest bet is something like:
- do full backup
- pause write load
- do incremental backup (should be quick compared to previous full backup)
- start a new cluster on 3.0.7, start with one instance on the backup-db, which will migrate the db from 2.3.1 --> 3.0.7
- start the other instances of the new 3.0.7 cluster, which will copy the db from the 3.0.7 master. (do this asynchronically, don't wait for them to complete the copy).
- redirect all load to new cluster
- resume write load
Mattias Finné
- 3,034
- 1
- 15
- 7
-
Thanks for your valuable input. on your step 2 "pause write load", this means we have downtime correct ? – Renjith Dec 28 '16 at 06:54
-
Sure although a matter of definition, reads will still be available – Mattias Finné Dec 29 '16 at 09:21
-
Is there any way that we can find the Delta, like mysql ? – Renjith Jan 01 '17 at 14:59