Add Update documentation page

diff --git a/doc/Home.md b/doc/Home.md
index ce23f3e17bc478d7cb69d4ebbd815e582445ca9d..4f956e3c7a8560bb6324fcf6ddc33f2f4924872b 100644 (file)
--- a/doc/Home.md
+++ b/doc/Home.md
@@ -27,6 +27,7 @@ Friendica Documentation and Resources
 **Admin Manual**
 
 * [Install](help/Install)
+* [Update](help/Update)
 * [Settings & Admin Panel](help/Settings)
 * [Installing Connectors (Twitter/GNU Social)](help/Installing-Connectors)
 * [Install an ejabberd server (XMPP chat) with synchronized credentials](help/install-ejabberd)

diff --git a/doc/Install.md b/doc/Install.md
index 10975d37ef583cb412efba30026ea5e88fd823cc..c9fc4e6031224c6d1138e7a4f7c81b60569dd797 100644 (file)
--- a/doc/Install.md
+++ b/doc/Install.md
@@ -118,24 +118,6 @@ If it is not possible to set up a cron job then please activate the "frontend wo
 
 Once you have installed Friendica and created an admin account as part of the process, you can access the admin panel of your installation and do most of the server wide configuration from there
 
-Updating your installation with git
----
-
-You can get the latest changes at any time with
-
-    cd mywebsite
-    git pull
-    util/composer.phar install
-
-The default branch to use it the ``master`` branch, which is the stable version of Friendica.
-If you want to use and test bleeding edge code please checkout the ``develop`` branch.
-The new features and fixes will be merged from ``develop`` into ``master`` when they are stable approx four times a year.
-
-The addon tree has to be updated separately like so:
-
-    cd mywebsite/addon
-    git pull
-
 ###Set up a backup plan
 Bad things will happen.
 Let there be a hardware failure, a corrupted database or whatever you can think of.

diff --git a/doc/Update.md b/doc/Update.md
new file mode 100644 (file)
index 0000000..57d29ba
--- /dev/null
+++ b/doc/Update.md
@@ -0,0 +1,72 @@
+Updating Friendica
+===============
+
+* [Home](help)
+
+## Using a Friendica archive
+
+If you installed Friendica in the ``path/to/friendica`` folder:
+1. Unpack the new Friendica archive in ``path/to/friendica_new``.
+2. Copy ``.htconfig.php``, ``photo/`` and ``proxy/`` from ``path/to/friendica`` to ``path/to/friendica_new``.
+3. Rename the ``path/to/friendica`` folder to ``path/to/friendica_old``.
+4. Rename the ``path/to/friendica_new`` folder to ``path/to/friendica``.
+5. Check your site. Note: it may go into maintenance mode to update the database schema.
+6. If everything works, just delete the ``path/to/friendica_old`` folder.
+
+To update Addons from an archive, simply delete the ``path/to/friendica/addon`` and replace it with the provided archive.
+
+## Using Git
+
+You can get the latest changes at any time with
+
+    cd path/to/friendica
+    git pull
+    util/composer.phar install
+
+The addon tree has to be updated separately like so:
+
+    cd path/to/friendica/addon
+    git pull
+
+For both repositories:
+The default branch to use is the ``master`` branch, which is the stable version of Friendica.
+It is updated about four times a year on a fixed schedule.
+
+If you want to use and test bleeding edge code please checkout the ``develop`` branch.
+The new features and fixes will be merged from ``develop`` into ``master`` after a release candidate period before each release.
+
+Warning: The ``develop`` branch is unstable, and breaks on average once a month for at most 24 hours until a patch is submitted and merged.
+Be sure to pull frequently if you choose the ``develop`` branch.
+
+### Considerations before upgrading Friendica
+
+#### MySQL >= 5.7.4
+
+Starting from MySQL version 5.7.4, the IGNORE keyword in ALTER TABLE statements is ignored.
+This prevents automatic table deduplication if a UNIQUE index is added to a Friendica table's structure.
+If a DB update fails for you while creating a UNIQUE index, make sure to manually deduplicate the table before trying the update again.
+
+#### Manual deduplication
+
+There are two main ways of doing it, either by manually removing the duplicates or by recreating the table.
+Manually removing the duplicates is usually faster if they're not too numerous.
+To manually remove the duplicates, you need to know the UNIQUE index columns available in `database.sql`.
+
+```SQL
+SELECT GROUP_CONCAT(id), <index columns>, count(*) as count FROM users
+GROUP BY <index columns> HAVING count >= 2;
+
+/* delete or merge duplicate from above query */;
+```
+
+If there are too many rows to handle manually, you can create a new table with the same structure as the table with duplicates and insert the existing content with INSERT IGNORE.
+To recreate the table you need to know the table structure available in `database.sql`.
+
+```SQL
+CREATE TABLE <table_name>_new <rest of the CREATE TABLE>;
+INSERT IGNORE INTO <table_name>_new SELECT * FROM <table_name>;
+DROP TABLE <table_name>;
+RENAME TABLE <table_name>_new TO <table_name>;
+```
+
+This method is slower overall, but it is better suited for large numbers of duplicates.
\ No newline at end of file