Reindex after DB restore

[Flavio Beck]

Hello.

I have created test instances many times along last years, basically like this guide

https://confluence.atlassian.com/adminjiraserver/restoring-information-from-a-native-backup-938847709.html

after restore all data (home and db) jira was used to start normally (like production) and reindex was not required.

This week I tried this, but jira is starting without any indexed issue.

JIRA

- jira prodution (7.10) is running as docker container.

backup and restore is just a rsync from jira_home directory

- database (postgres 9.5) docker container

I tried use pg_dump   and also pg_dumpall -> same result (no indexed issues)

Test:

I connected test instance to production DB and it worked perfectly (all issues found without reindex)

but

connecting test instance to a restored db (test) does not work, (no issues found)

any idea?

could be related to this workaround done?

https://jira.atlassian.com/browse/JRASERVER-63002

[Flavio Beck · Accepted solution]

I've found the problem.

I use rsync to copy the home directory  (from production to test env), for some reason when running rsync while production is up, when I start the test server It identifies "file not found" in cache directory and clean up all indexes, so it requires re-index from scratch.

but,  running rsync after stop the production server, the test server starts with all indexes working fine. (reindex is not required)

It make sense, but I have done this many times along last years without stop production, and never lost index files before... 

anyway, it is solved.

thanks

[Fabio Racobaldo _Herzum_]

Hi @Flavio Beck ,

Are issues not visible just using filters? If yes, run a reindex. My suggestion is to perform a reindexing from scratch (manually delete index folder from the server) because the process is quicker.

Regrads,

Fabio

[Flavio Beck]

yes, reindex fix the problem, but I was really trying to understand what happened.

Today I cloned the DB VM and I connected the Jira (test) to this DB, and again, there are no indexed issues.

I mean, if I go to search, there is no results (it should have more than 2 millions issues)

If I index any project, so I can see that in search results.

But note this scenario..

DB -> it is a clone VM from production

JIRA - >  it runs as docker container in production (and for test as well), and I copied the home directory from production volume (it is the docker persistency volume). 

Same docker image of production.

it is the similar if I was trying to recreate the production container and lost the index.

(same DB and same persisted home volume)

it does not make sense to lost the index.