Cannot create a backup from older version of JIRA.

[Pat B.]

We've been upgrading Jira from an older release to the actual 7.8.1, hosted on Windows Server 2008R2 on MSSQL.

After performing this, we're having some strange issues.

1. Cannot start a manual XML backup

   Fehler

   • Ausnahme bei Sicherung: java.lang.IllegalStateException: TABLE: AO_013613_EXPENSE: AMOUNT - CREATED - CREATED_BY - DATE - DESCRIPTION - EXPENSE_CATEGORY_ID - ID - SCOPE - SCOPE_TYPE - can't find type -151 (precision=23,scale=3) in field CREATED
2. Cannot create a Jira Help Desk project
   1. Hmm... we couldn't create your project due to an unknown error. Try refreshing the page to start again

How can we correct this? It's the first time we're having Issues with Jira - Please tell me how to troubleshoot this.

We will then move the Jira installation to a Ubuntu/Postgres based setup.

Kind regards

patrick

[bin liu · Accepted solution]

Maybe it is the's jira implemention, I have moved sub ou to the same user directory, and access speed is faster than previous.

[DanielP]

Of course it's faster. Now Jira will only do 1 request to the AD rather than 20.

[bin liu]

The first user directory and the twentieth user directory all are sub "ou" of the same active directory. I think they will take the same time when calling ldap server to authenticate.

[DanielP]

But in order to find a user in the twentieth directory Jira will have to check ALL THE OTHER 19 FIRST. It doesn't matter if the directories are all sub "ou" of the same AD. Jira has no way of in wich directory a user exist before it has checked all the directories.

Let me illustrate:
Imagine yourself standing before a line of 20 people. You have to find the person named Steve. To do this you will have to ask each person in the line for their name, one after the other.

You: "What's your name?"
Person 1: "John"
You: "What's your name?"
Person 2: "Andy"
...
And so on until you find Steve.

If Steve is the last person in the line it will take a lot of time to find him. If he on the other hand is the first in line it will go rather quickly.

The same principle applies to Jira's user directories.

[bin liu]

finding a member of the 20th directory is very fast when I use SQL, I don't understand why does jira take too long time to search a member.

sql like below:

select user_name, directory_id, display_name, email_address, dir.directory_position as position

from jirapdb.cwd_user usr

join jirapdb.cwd_directory dir on dir.id = usr.directory_id

where user_name = 'scmercwd_group'

order by dir.directory_position

[Timothy]

Can you confirm that the code uses this SQL to search through the directories?

[DanielP]

Finding the user in Jira's (or crowd's) database isn't what takes time. It's actually calling all of the ldap servers to authenticate that takes time.

This will also be heavily dependent of the exact type of user directory you are using. Using "Internal with LDAP authentication" will probably be quicker than using pure LDAP directories.

[DanielP]

Hi,

The delay comes from the fact that Jira actually tries all the directories in sequence until it finds the user it's looking for. So if your are a member of the 20th directory, Jira will query all the other 19 directories first. This will inevetably take a longer time than if you are a member of the first directory. The only way of speeding things up is as far as I know to cut down on the number of directories you are using.