Migration from Server to DC and changing the database from MYSQL to PostgreSQL

[Marco Göhl]

Hello Migration Community,

we are currently on the way to migrate jira server to jira dc.
Doing this we want to change the database from MySQL to PostresSQL.

Our journey is running for some time now, and we currently stand still more or less.
We do the migration by taking an export from the source system and then try to import it on the target system.
We run into an amount of issues which we fix step by step, sometimes with the help of the atlassian support.

In the moment the import stuck at 90% and breaks...and breaks. At 90% there starts the part to takeover the AO tables/data, which have an dependency to the jira plugins.

There are two things i'm intrested in:
1st - does anybody else have experience doing this and is it the right way starting the import again and again, fixing failure by failure?
In a huge environment this is really time consuming.

2nd - we are facing data incosistencies currently, especially for sla audit tables. We understand that the system is inserting/updating while the backup is running, which could be the reason for it.
For my understanding it is then a lucky punch to get the restore to the end. Can someone tell something around this?

Kind regards
Marco

[Mark Benson]

I've done the migration between server and DC
Then from DC to Cloud

None of them were fun or easy... and full of problems similar to what you're describing.

___________

Whenever we performed an upgrade to our DC cluster version I'd always dry run test this as it was also often prone to issues upgrading or would break other things like addons etc.

Even doing a DC to DC restore same version both sides (cloning the enviro) I'd very frequently run into AOTables related errors after the restore completed which would break certain JSM functionality.

This is several years ago now so it's not terribly fresh in my memory but I'd need to CLI into a Linux  cluster node, connect to the database using psql and manually massage the database and increment +1 for each erroring AOTables row.

Sadly there's so many different kinds of errors related to XML restores so it's hard to really say where yours is bombing without seeing the exact logs.

If you Google "jira restore aotables" you'll find a whole bunch.. but based on your description it sounds like this could be the one affecting you:

https://jira.atlassian.com/browse/JRASERVER-70147

I have had to do this in the past as well to bypass an issue, spin up a fresh instance of Jira Software, perform the restore, then add the JSD (now JSM) component after the instance is up and running.

Good luck!

[Marco Göhl]

Hi Mark,

thanks for your reply!
Seem the error with JSD is solved in higher version and atlassian is advising installing it before restoring in the same version as on source.

We will do a next try with deleting the AO_...SLAAUDIT Tables before creating a backup.

Lets see if this will bring us a bit forward.

But your comment tells me that it is always a hard way to go!

Kind regards
Marco

[Vijayraj Amin]

Hi Marco,

We have 20+ years hands on experience in Data Migration and Data Transformation. YES, we can help you with your immediate requirement. 

Write back to vijayraj.amin@maiora.co to discuss in detail.

Thank you! 

[Marco Göhl]

Hi Vijayrai,

thanks for your reply!

We follow the atlassians supported way for this migration, which is backup/restore via xml. Its not only about shifting the data between two databases.

But in the past we gave that already a try with AWS DMT Tools, but at the end we were also facing failurs. As it was an unsupported way, we can not ask the support.

I'm not sure which way of migrating the data you are talking about.

Kind regards
Marco 

[Kirana AB]

Hi Marco Göhl,

In regards to a change the Jira database, I would suggest trying to follow the steps:https://confluence.atlassian.com/adminjiraserver/switching-databases-938846867.html. 

This process has you creating an XML backup .zip file in Jira and then creating a new blank database for Jira to use.  This way you can then import that data using Jira's setup wizard to import that backup into this new empty database.

[Marco Göhl]

Hello Kirana,

Thanks for your reply!

We are aware of the things you mentioned, and we followed it.
But seem to be a bit normal to have this kind of problems and there is a need to fix it step by step.

Kind regards
Marco