Difference between revisions of "Tips for large databases"

From Gramps
Jump to: navigation, search
m (Avoid general filters)
(Avoid relationship calculation)
(22 intermediate revisions by 6 users not shown)
Line 1: Line 1:
Large data files, what to do, and what not to do.
+
{{out of date|check accuracy}}
 +
 
 +
Large family tree data files, what to do, and what not to do.
  
 
==Loading the file==
 
==Loading the file==
[[Image:EditPreferencesGeneral226.png|thumb|General tab of Edit->Preferences dialog box]]
 
For small files, you can work on GEDCOM or GRAMPS XML format. This works great.
 
However, these formats are completely loaded in your computer memory so, for larger files, this becomes slow and might bring your computer to a standstill.
 
  
Therefore, you should only handle large files by using the GRAMPS GRDB format. You do this by creating a new GRDB file and importing the large file into it. This can take a long time, so do it in the evenings! If you '''enable database transactions''' in the  Edit->Preferences->General menu tab, the importing of data will then be much faster.
+
Initial import of a large (100,000+) database from either Gramps XML formats or GEDCOM is can take a few hours. You will need to adjust the
 +
number of allowable locks. For 140,000 people you should use:
 +
* ''max_locks'' 300000
 +
* ''max_objects'' 300000
  
Also if you have a file that large it might be prudent to break up file into smaller pieces (i.e. Maternal & Paternal lines in different files, if they are not interrelated) so you only have to use a smaller file.
+
The easiest way to do this is to:
<br clear="all">
+
# create a new Family Tree
 +
# exit Gramps
 +
# add a '''DB_CONFIG''' file to the database directory before importing (see '''gramps -l''' output for the directory of a specific family tree.)
 +
 
 +
Contents of this DB_CONFIG file should be:
 +
#may want to fiddle with cachesize also
 +
#set_cachesize 0 200000000 2
 +
set_lk_max_locks 300000
 +
set_lk_max_objects 300000
 +
 
 +
{{-}}
  
 
== Default view: bookmarks and active person ==
 
== Default view: bookmarks and active person ==
As there are many people, loading the person view might take some time. See a comparison of [[GRAMPS Performance|performance]] on large datasets, between different GRAMPS versions.
 
  
You can circumvent this load time by going to the [[Relationships screenshot|relationships view]] before opening the grdb file. This will open the active person and their family extremely fast, independent of the amount of data you gathered.
+
As there are many people, loading the person view might take some time. See a comparison of [[Gramps Performance|performance]] on large datasets, between different Gramps versions.
 +
 
 +
You can circumvent this load time by going to the [[Relationships screenshot|relationships view]] before opening the family tree. This will open the active person and their family extremely fast, independent of the amount of data you gathered.
  
 
You can add bookmarks to common people, or in the branch you are researching. This allows to change person in the relationships view without the need to activate the person in the person view first.
 
You can add bookmarks to common people, or in the branch you are researching. This allows to change person in the relationships view without the need to activate the person in the person view first.
 +
 +
== Avoid relationship calculation ==
 +
 +
Avoid displaying relationships on the status bar. From the {{man menu|Edit > Preferences}} ''Display'' tab change the '''Status bar''' option from {{man label|Relationship to the home person}} to {{man label|Active person's name and ID}}.
 +
 +
== Avoid Gramplets ==
 +
Avoid gramplets which do a lot of database work. The '''[[Deep_Connections_Gramplet|Deep Connections]]''' Gramplet seems to be the worst case. These gramplets slow everything down enormously.
  
 
== Avoid general filters ==
 
== Avoid general filters ==
 +
 
Filters search all the data looking for the people who satisfy your query. Try to avoid this on large files.
 
Filters search all the data looking for the people who satisfy your query. Try to avoid this on large files.
  
{{man tip|Search letters|Fast search can be done by enabling column search (disable {{man label|Sidebar}} on {{man label|View}} menu), but this will search all letters on words into the column, not only the first letters.}}
+
{{man tip|Search letters|Fast search can be done by enabling column search (disable {{man label|Sidebar}} on {{man menu|View}} menu), but this will search all letters on words into the column, not only the first letters.}}
  
 
For a surname search:
 
For a surname search:
 
* click  on a person in the [[People screenshot|people view]]. Now start typing the family name of the person you need. The view will automatically go to the family name as you complete your typing.
 
* click  on a person in the [[People screenshot|people view]]. Now start typing the family name of the person you need. The view will automatically go to the family name as you complete your typing.
* for a more detailed search, click ''CTRL-F'' first. This will expand all the nodes and allows to search more specifically.
+
* for a more detailed search, use the keyboard shortcut {{man key press|CTRL|F}}. This will expand all the nodes and allows to search more specifically.
  
 
== Flat views are faster than Tree views ==
 
== Flat views are faster than Tree views ==
  
Using large databases and person treeview needs to often rebuild the view.
+
Using a large database and person treeview needs to often rebuild the view.
 +
 
 
Flat person view (list) is faster.
 
Flat person view (list) is faster.
 
== Please contribute ==
 
If you have more tips for working with large databases, please share them here.
 
  
 
[[Category:Documentation]]
 
[[Category:Documentation]]

Revision as of 03:47, 17 July 2015

Gramps-notes.png This page's factual accuracy may be compromised due to out-of-date information. Please help improve the Gramps Wiki as a useful resource by updating it.

Large family tree data files, what to do, and what not to do.

Loading the file

Initial import of a large (100,000+) database from either Gramps XML formats or GEDCOM is can take a few hours. You will need to adjust the number of allowable locks. For 140,000 people you should use:

  • max_locks 300000
  • max_objects 300000

The easiest way to do this is to:

  1. create a new Family Tree
  2. exit Gramps
  3. add a DB_CONFIG file to the database directory before importing (see gramps -l output for the directory of a specific family tree.)

Contents of this DB_CONFIG file should be:

#may want to fiddle with cachesize also
#set_cachesize 0 200000000 2
set_lk_max_locks 300000
set_lk_max_objects 300000


Default view: bookmarks and active person

As there are many people, loading the person view might take some time. See a comparison of performance on large datasets, between different Gramps versions.

You can circumvent this load time by going to the relationships view before opening the family tree. This will open the active person and their family extremely fast, independent of the amount of data you gathered.

You can add bookmarks to common people, or in the branch you are researching. This allows to change person in the relationships view without the need to activate the person in the person view first.

Avoid relationship calculation

Avoid displaying relationships on the status bar. From the Edit > Preferences Display tab change the Status bar option from Relationship to the home person to Active person's name and ID.

Avoid Gramplets

Avoid gramplets which do a lot of database work. The Deep Connections Gramplet seems to be the worst case. These gramplets slow everything down enormously.

Avoid general filters

Filters search all the data looking for the people who satisfy your query. Try to avoid this on large files.

Tango-Dialog-information.png
Search letters

Fast search can be done by enabling column search (disable Sidebar on View menu), but this will search all letters on words into the column, not only the first letters.


For a surname search:

  • click on a person in the people view. Now start typing the family name of the person you need. The view will automatically go to the family name as you complete your typing.
  • for a more detailed search, use the keyboard shortcut CTRL+F. This will expand all the nodes and allows to search more specifically.

Flat views are faster than Tree views

Using a large database and person treeview needs to often rebuild the view.

Flat person view (list) is faster.