TRIP - Trip Recording and Itinerary Planner

Introduction

TRIP is a JavaScript web-based application supporting trip recording and itinerary planning. It is designed to be a lightweight server application such that it can be run on relatively low power devices like the Raspberry Pi.

The intended use is for a hiker, mountain-biker or other adventurer, to be able to publish and share their planned itinerary, then subsequently log their positions at intervals, allowing someone else to be able to monitor their progress.

In the event of contact being lost, the plans and tracking information can be passed to rescue services etc., to assist with locating the missing adventurer.

The web client is an AngularJS single-page application (SPA) which can be served as static files using a web server such as Apache, or served by the TRIP server application itself.

The following features are provided:

  • Remote tracking server—client applications such as TripLogger Remote for iOS – (on the App Store) or GPSLogger for Android can be used to submit locations to the server.

  • Sharing tracks with others.

  • Viewing tracks on a map provided by a tile server, e.g. OpenStreetMap tiles.

  • Creating and sharing itineraries using the Markdown markup language.

  • Using the map, interactively creating routes and waypoints for an itinerary.

  • Uploading and downloading routes, tracks and waypoints for an itinerary as a GPX file.

  • Viewing routes, tracks and waypoints of an itinerary on the map.

  • Splitting and joining routes and tracks.

  • Deleting individual points from routes and tracks.

Requirements

  • Node.js - v10.x.x

  • PostgreSQL database server - (Known to run on version 11.9)

  • PostGIS PostGIS spatial extension to PostgreSQL (2.5.x)

Quick Start

You can quickly get the full application up and running either using Docker or Vagrant virtualisation systems. Docker is probably the simplest and easiest method to use. Using both are described in following sections.

Quick Start Using Play with Docker

Use Play with Docker to run the application with some test data in a browser, without having to install anything.

Navigate to Play with Docker and login using a Docker ID. If you do not have one, you will see the option to sign up after clicking Login then Docker.

Click + ADD NEW INSTANCE

Cut and paste each of the following commands in sequence into the terminal window. This will create a Docker network and run two containers, one providing the database and the other the application web server.

$ docker network create trip-server

$ docker run --network trip-server --network-alias postgis \
-e POSTGRES_PASSWORD=secret -d fdean/trip-database

$ docker run --network trip-server -e TRIP_SIGNING_KEY=secret \
-e TRIP_RESOURCE_SIGNING_KEY=secret -e POSTGRES_PASSWORD=secret \
--publish 8080:8080 -d fdean/trip-server

Once the application is running a link titled 8080 will be shown next to the OPEN PORT button at the top of the page. Click on the 8080 link to open a new browser window to the running web server.

Login using one of the following users and credentials:

user@trip.test  rasHuthlutcew7
admin@trip.test 7TwilfOrucFeug

See the user documentation for information on using the application.

Quick Start Using Docker

  1. Follow the instructions for installing a Docker environment.

  2. Clone this repository with git.

  3. Run the following command in the root of the cloned repository:

    $ sudo docker-compose up -d
    
  4. Navigate to http://localhost:8080/app with a web-browser. The application runs with a small amount of test data.

  5. Login using one of the following users and credentials:

    user@trip.test  rasHuthlutcew7
    admin@trip.test 7TwilfOrucFeug
    
  6. Click Help in the Trip Web Client menu for the online user documentation.

The Docker container can be stopped with:

$ sudo docker-compose down

The script creates a Docker volume named 'trip-server_trip-db-data' for the database store. List the Docker volumes with:

$ sudo docker volume ls

If you stop the container with the --volumes option, the database volume is removed as well as the container:

$ sudo docker-compose down --volumes

Alternatively, to remove the volume after the container has been stopped, it can be removed with:

$ sudo docker volume rm trip-server_trip-db-data

To use a Docker container for development:

$ sudo docker-compose --file docker-compose-dev.yml up --build -d
$ sudo docker-compose logs --follow

The environment requires both the trip-server and trip-web-client projects to share the same parent folder. These folders are mounted within the Docker container such that changes made on the host are also reflected within the container.

The server will automatically restart if any of it's .js or .json files are altered.

When making changes to the trip-web-client HTML or JavaScript, you will need to refresh the browser to replace the cached version.

Stop the Docker containers with:

$ sudo docker-compose --file docker-compose-dev.yml down

Optionally, add the --volumes parameter to remove the Docker volumes when shutting down.

Alternatively, the Docker volumes can be removed with:

$ sudo docker volume rm trip-server_node_modules \
trip-server_trip-web-client trip-server_web_node_modules \
trip-server_trip-db-data

Docker Swarm

The source code also contains a configuration file named docker-compose-swarm.yml for running the application as a Docker swarm. This uses Docker secrets instead of environment variables to pass secrets to the container. This can be used as the basis for a production configuration, but it does not work across multiple nodes as it is using Docker's local volume storage provider. A third-party swarm aware storage provider is required to work across multiple nodes.

After initialising the swarm, the secrets can be created as follows:

$ echo 'secret' | sudo docker secret create postgres_password -
$ echo 'secret' | sudo docker secret create jwt_signing_key -
$ echo 'secret' | sudo docker secret create jwt_resource_signing_key -

Note that the postgres_password is used to form a URI so must not contain a forward-slash character.

See DockerTips which contains some notes on running a Docker swarm.

Quick Start Using Vagrant

This option provides a working example of the application running in a VirtualBox virtual machine (VM) for those operating systems supported by Vagrant. This also provides a complete example of running the application behind the Nginx ("engine x") HTTP reverse proxy server. It is suitable for development or demonstration, but not as a production system.

Note: Installing all the required software, including the Vagrant box involves downloading approximately 600MB of data. Perhaps more of an "easy-start" rather than a "quick-start".

  1. Download and install VirtualBox

  2. Download and install Vagrant

  3. Clone this repository to a suitable location on the machine you are going to use to host the application and VM:

    $ cd ~/projects
    $ git clone git://www.fdsd.co.uk/trip-server.git
    
  4. Clone the TRIP web client application to have the same parent folder as the TRIP server

    $ cd ~/projects
    $ git clone git://www.fdsd.co.uk/trip-web-client.git
    
  5. Start the Vagrant VM

    $ cd ~/projects/trip-server
    $ vagrant up
    

    The first time this is run, it will download a Vagrant box containing a Debian Linux distribution, then install the required Debian packages, modify the default configuration and start the TRIP server running behind an Nginx web-server

  6. Make a note of the trip-server admin user credentials displayed at the end of the startup process

  7. Use your browser to navigate to http://localhost:8080/ on the host machine and login providing the above credentials

  8. When finished, halt the server with:

    $ vagrant halt
    

Vagrant shares the two source folders with the VM so that you can modify the source files on the host server and immediately impact the deployed application. This gives you a complete working development environment.

Note: When the VM is initially provisioned, if the node_modules sub-folder exists, it is removed and the contents re-installed/rebuilt to ensure that all of the yarn installed binaries are compatible with the VM's operating system. Similarly, you may need to remove and re-create this folder if you choose to run the trip server directly on the local guest machine.

Rendering of map tiles is disabled by default, in order to respect OpenStreetMap's Tile Usage Policy. You will need to follow the instructions below, in the Tile Server Configuration section, before map tiles are rendered.

If you forget the admin user (admin@trip.test) password, login into the VM and modify the database entry in the PostgreSQL database. Replace SECRET with your desired password.

$ cd ~/projects/trip-server
$ vagrant ssh
$ psql trip
trip=# UPDATE usertable SET password=crypt('SECRET', gen_salt('bf')) WHERE nickname='admin';
trip=# \q

You can configure the time zone and locale settings by running the following commands on the guest VM and following the prompts:

$ sudo dpkg-reconfigure tzdata
$ sudo dpkg-reconfigure locales

Optionally, apply the latest Debian updates with:

$ sudo apt-get upgrade

View the Vagrantfile configuration file in the root of the trip-server folder for some examples you can modify. E.g. you can enable the config.vm.network "public_network" option to make the VM accessible from the public network. This would allow you, for example, to test location updates, using a GPS enabled device sharing the same private LAN as the host VM. Note the warnings in the Vagrant documentation for this setting, as for convenience, the VM is insecure by default and design.

Trouble-shooting

Guest additions on this VM do not match the installed version of VirtualBox!

This means the installed box needs updating or an older version of VirtualBox needs to be used.

A simple solution is to install the vagrant-vbguest package.

    $ vagrant plugin install vagrant-vbguest

See also https://stackoverflow.com/questions/20308794/how-to-upgrade-to-virtualbox-guest-additions-on-vm-box

Vagrant has detected a configuration issue which exposes a vulnerability with the installed version of VirtualBox

e.g.

    Vagrant has detected a configuration issue which exposes a
    vulnerability with the installed version of VirtualBox. The
    current guest is configured to use an E1000 NIC type for a
    network adapter which is vulnerable in this version of VirtualBox.
    Ensure the guest is trusted to use this configuration or update
    the NIC type using one of the methods below:

      https://www.vagrantup.com/docs/virtualbox/configuration.html#default-nic-type
      https://www.vagrantup.com/docs/virtualbox/networking.html#virtualbox-nic-type

Should be fixed in VirtualBox 5.2.22, but error still reported by version 2.2.22 of Vagrant. See https://github.com/hashicorp/vagrant/issues/10481

Standard Setup and Configuration

This section describes manually installing and configuring a server to run the TRIP application. It can be run as a standalone Node.js server, or behind a reverse proxy server, such as Apache or Nginx.

These instructions assume installation on a Debian based Linux system, but it should run on any system supported by Node.js.

The basic installation consists of downloading and configuring the trip-server application, the trip-web-client application, and configuring with the PostgreSQL database server.

On a Debian 10 (Buster) system, install the following packages:

    $ sudo apt-get install postgresql postgresql-contrib postgis

The following package will be installed automatically, unless you have set APT::Install-Recommends to false in apt preferences. If they aren't automatically installed:

    $ sudo apt-get install postgresql-11-postgis-2.5 postgresql-11-postgis-2.5-scripts

If the application is exposed to the Internet, ideally it should also be configured to run behind an Apache web server using HTTPS.

  1. Install Node.js. See Installing Node on Linux.

  2. Install yarn

  3. Download the trip-server application and install it in an appropriate folder, e.g. /usr/local/trip-server

  4. Install the packages required by trip-server:

    $ cd /usr/local/trip-server
    $ yarn install
    

    If bcrypt fails to build, you probably need to install C++ build tools etc. In Debian this is most easily achieved by installing the build-essential package.

  5. Clone the trip-web-client to an appropriate folder, ideally outside the server's folder, with a symlink from the app sub-directory of the server to the client's top-level folder. e.g. /usr/local/trip-web-client.

    $ cd /usr/local/trip-server
    $ ln -s ../trip-web-client/app
    

    On systems that do not support symlinks, it can be installed in the app sub-folder for running as a standalone development server.

  6. Install the packages required by trip-web-client:

    $ cd /usr/local/trip-server/app
    $ yarn install
    
  7. TRIP server's configuration is maintained in a file named config.json or config.yaml in the application's root directory. Create either the JSON or YAML format according to your personal preference. If you don't have a preference, YAML is probably easier to use. If both exist, config.yaml is chosen in preference by TRIP.

    Create the initial version by making a copy of config-dist.json or config-dist.yaml and modifying to suit your environment and preferences.

    Make sure the file is not world-readable as it will contain the database and token signing passwords. Review the file, modifying entries to suit your requirements.

    • app.json.indent.level - Indent level when debugging server with pretty print enabled

    • app.origins - Valid origins to handle Cross-Origin Resource Sharing (CORS). Typically this is the protocol, host domain and port, e.g. https://example.com:443. The default is *:* which currently allows all origins.

    • jwt.signingKey - Create a strong password to sign authentication tokens with

    • tile.cache.maxAge - The number of days to cache tiles for. They will always be cached for at least the minimum set by the tile server's expires header. This setting allows that period to be increased.

    • tile.providers - These are arrays which configure the remote tile server for maps. (See Tile Server Configuration below).

    • db.uri - the URI for the trip user to connect to the trip database. Replace the word 'secret' with the database password

    • staticFiles.allow - true to serve trip-web-client static files, false if you are serving them from elsewhere, e.g. Apache

    • debug - true provide error message JSON objects in responses and log all requests - should be false in a production environment

    • log.level - Sets the level of server logging - one of, debug, info, warn or error

    • reporting.metrics.tile.count.frequency - How frequently to capture the number of request that have been made to the remote tile server

Tile Server Configuration

Most if not all tile server providers have policies that you must comply with and there may be sanctions if you fail to do so. E.g. If you are using the OpenStreetMap tile server, read and comply with their Tile Usage Policy. Please ensure you configure the following entries correctly for the appropriate element of the tile.providers section(s) of config.json or config.yaml.

  • userAgentInfo - This is the e-mail address at which the system administrators can contact you

  • refererInfo - A link to a public website with information about your application's deployment

Note these entries are sent in the HTTP header of each tile request and will therefore end up in system logs etc. Currently the tile requests are sent over HTTP, therefore you should not mind this data being exposed.

The tile.providers[x].mapLayer entries provide the ability to display tile map attributions most if not all tile providers require you to display.

The mapLayer.name attribute will be displayed when the map layers icon is activated. Only xyz map types are supported, so the mapLayer.type attribute should always be xyz.

Map attributions are displayed on the map using the mapLayer.tileAttributions section of the tile.providers attribute, which allows attributions to be rendered with appropriate HTML links. The tileAttributions are an array of items that have either text, text and link or just link attributes. If the entry contains just text, the text will be displayed in the map attribution. If a link is included, the text will be wrapped in HTML link tags and included in the map attribution. The entries are displayed in the sequence they have been defined.

Elevation Data

The Consortium for Spatial Information (CGIAR CSI) make Digital Elevation Model data covering about 80% of the globe, available for download. It has been sourced and enhanced from data gathered by the NASA Shuttle Radar Topographic Mission (SRTM).

From the main page of the CGIAR CSI website, follow the link to SRTM Data to download zip files that contain tiff files with 5m x 5m elevation data.

Extract the tiff files to a folder, e.g. /var/local/elevation-data and configure an elevation section in config.json or config.yaml, e.g.

...

    "elevation" : {
        "tileCacheMs" : 60000,
        "datasetDir" : "/var/local/elevation-data/"
    },

...

When the Trip Server application is started, it reads all the tiff files in the folder specified by the elevation.datasetDir parameter and creates an in memory index containing the area covered by each tile. When elevation data is required for a specific location, the relevant tile is loaded, the response provided, and the tile retained in memory for the number of milliseconds specified by the elevation.tileCacheMs parameter.

The tiff files take up a lot of space. Where space is at a premium, consider storing them in a compressed file system, e.g. on Linux use Squashfs.

e.g.

  1. Download files to ~/downloads/srtm

    $ mkdir -p ~/downloads/srtm
    $ cd ~/downloads/srtm
    $ wget http://srtm.csi.cgiar.org/wp-content/uploads/files/srtm_5x5/tiff/srtm_72_22.zip
    
  2. Extract the tiff files to ~/tmp/tiff

    $ mkdir -p ~/tmp/tiff
    $ cd ~/tmp/tiff
    $ find ~/downlods/srtm -name '*.zip' -exec unzip -n '{}' '*.tif' \;
    
  3. Create a Squashfs compressed file containing the tiff images

    $ mksquashfs ~/tmp/tiff /var/local/elevation-data.squashfs -no-recovery
    

    The -no-recovery option is to stop Squashfs leaving a recovery file behind in the destination folder. However, it does mean that should the operation fail, there is no recovery information to unwind the command. This is probably more of a potential problem when appending to an existing Squashfs file.

  4. Optionally, delete or archive the downloaded zip files to free up space.

  5. Download more files, extract them and squash them using the above steps. Repeating the mksquashfs command as above will append to an existing Squashfs file.

  6. You can list the contents of the Squashfs file with:

    $ unsquashfs -i -ll /var/local/elevation-data.squashfs
    
  7. Test mounting the Squashfs file

    $ mkdir -p /var/local/elevation-data/
    $ sudo mount -t squashfs /var/local/elevation-data.squashfs /var/local/elevation-data/
    $ ls /var/local/elevation-data/
    $ sudo umount /var/local/elevation-data
    
  8. Add an entry to /etc/fstab to mount the Squashfs file on boot:

    $ echo '/var/local/elevation-data.squashfs /var/local/elevation-data squashfs ro,defaults 0 0' \
      | sudo tee -a /etc/fstab
    
  9. Mount using the /etc/fstab entry:

    $ sudo mount /var/local/elevation-data
    $ ls /var/local/elevation-data
    $ sudo umount /var/local/elevation-data
    
  10. If need be in the future, you can extract the files from the Squashfs file with:

    $ unsquashfs -i /var/local/elevation-data.squashfs
    

    Which will extract all the files to a sub-folder of the current working folder named squashfs-root.

    Use the -f parameter if the squashfs-root folder already exists.

  11. To extract select files, create another file containing the names of the files to be extracted, prefixed by a forward-slash. e.g. /srtm_11_03.tiff.

    $ unsquashfs -i -e list-of-files.txt /var/local/elevation-data.squashfs
    

See SquashFS HOWTO for more information

PostgreSQL Database Configuration

Configure md5 password access for trip user to trip database

Add the following entry to /etc/postgresql/11/main/pg_hba.conf, just before the first entry for the all DATABASE and all USER type:

    local   trip     trip                             md5

Reload the database server configuration:

    $ sudo /etc/init.d/postgresql restart

or if using systemctl to manage daemons:

    $ sudo systemctl reload postgresql

Create the Database User

As a Unix user who is also a postgresql superuser:

    $ createuser -PDRS trip

This will command will prompt for a password for the user. This needs to match the password embedded in the db.url attribute in config.json or config.yaml when the application is deployed.

Create the Database

As a Unix user who is also a postgresql superuser:

    $ createdb --owner=trip trip

Confirm the trip user can connect to the database using the password created earlier:

    $ psql -d trip -U trip

Create tables and roles

As a Unix user who is also a postgresql superuser:

    $ cd ./spec/support
    $ psql trip <10_trip_role.sql
    $ psql trip <20_schema.sql
    $ psql trip <30_permissions.sql

Optionally, populate the database with data that can be used to perform end-to-end tests. Do not insert the test data into a production database as it contains default application admin user credentials.

    $ psql trip <90_test-data.sql

Lookup Tables

The following tables are used to define lookup values for select boxes in the web application:

  • waypoint_symbol - Key-value pairs describing waypoint symbols. The key is written to waypoint entries when downloading GPX files.

  • track_color - Key-value pairs together with an HTML color code. The 'key' is written to track entries when downloading GPX files and the HTML color code is used to render the tracks on the itinerary map page.

  • georef_format - Key-value pairs define how to format output of latitude and longitude values on the itinerary waypoint edit page. Format parameters are defined using the % symbol and have the following meanings:

    • %d - degrees
    • %m - minutes
    • %s - seconds
    • %D - zero prefix single digit degree values
    • %M - zero prefix single digit minute values
    • %S - zero prefix single digit second values
    • %c - output the cardinal value, S, E, W or N
    • %i - output a minus sign for W and S
    • %p - output a minus sing for W and S and a plus sign for E and N

E.g. a format string of %d°%M′%S″%c would result in a lat/long value of 1.5,-2.5 being displayed as 1°30′00″N 2°30′00″W

Scripts to create default values for these lookup tables are in the ./spec/support folder, named 60_waypoint_symbols.sql, 40_path_colors.sql and 50_georef_formats.sql respectively. The default waypoint symbols and track colours are generally appropriate for Garmin devices. If fact, the colours are the only ones allowed by the Garmin Extensions XSD.

Indexes for Query Performance

The location table has an index that is clustered on the time column to improve the query performance of date range queries. If the table becomes large and performance degrades, run the psql cluster command from time-to-time to re-cluster it. Note an exclusive lock is placed on the table for the duration of the cluster command execution.

See http://dba.stackexchange.com/questions/39589/optimizing-queries-on-a-range-of-timestamps-two-columns for more information.

Creating an Initial Admin User

An initial admin user needs to be created in the database. Thereafter, that user maintains other users using web application. Creating the initial admin user fundamentally consists of making entries in the usertable, role and user_role tables.

Firstly, create the entries in the role table by running the following script using psql:

    INSERT INTO role (name) VALUES ('Admin'), ('User');

There is a Node.js helper script (./spec/support/startup_helper.js) which will output the SQL commands to create an admin user. You can either output the result to a file and then run the file into psql, or pipe the output directly to psql. e.g.

    $ cd ./spec/support
    $ node startup_helper.js | psql trip

Standalone Node.js Server

The server can be run as a standalone server for development etc.

Run a terminal window and change to the directory where trip-server is installed.

    $ cd /usr/local/trip-server

The app sub-folder of this directory must contain the trip-web-client application, or be a symlink pointing to it. The trip-server application serves the trip-web-client files and other resources to the browser.

Start the server:

    $ node index.js

Using a browser, navigate to http://localhost:8080/app/index.html. The application should load in the browser and prompt for login.

The server can be stoped using ctrl-c in the terminal window.

Nginx Web Server

The application can be deployed with either the Apache or Nginx web servers. The section below this section describes setting up Apache.

Setting up Nginx isn't documented here but can readily be determined by looking at the Vagrant setup scripts under the ./provisioning/ folder in the source distribution, or by deploying using Vagrant and examining the working Vagrant installation.

Apache

Optionally, the application can be run behind an Apache server, proxying requests to the application.

This has the benefit of allowing the application to co-exist with other applications on the same server instance all running on the standard port 80. Security of the server can also be enhanced by installing and configuring the mod-security Apache module.

Reverse Proxy Configuration

Configure Apache2 to enable the mod_proxy and proxy_wstunnels modules. On Debian this can be done with:

    $ sudo a2enmod proxy
    $ sudo a2enmod proxy_wstunnel
    $ sudo a2enmod rewrite

The application must be run over HTTPS to keep the login credentials secure, otherwise others can see and re-use those credentials.

Modify the server configuration to implement the following rewrite rules. Note that the default socket.io path is prefixed with wstrack\ so that multiple applications using websockets can be run on the same Apache server. (TRIP uses websockets to provide updates to the tracking map.) The TRIP web client app will prefix the path when it is not calling a localhost URL. These rules need to be in a <VirtualHost _default_:443\> or <Directory\> section of the mod_ssl configuration file.

    RewriteEngine on
    RewriteCond %{REQUEST_URI}  ^/wstrack/socket.io  [NC]
    RewriteCond %{QUERY_STRING} transport=websocket    [NC]
    RewriteRule /wstrack/(.*)           ws://localhost:8080/$1 [P,L]

Add the following to trip.conf outside the <directory\> directive:

    <IfModule mod_proxy.c>
      ProxyPass /wstrack/socket.io/ http://localhost:8080/socket.io/
      ProxyPassReverse /wstrack/socket.io/ http://localhost:8080/socket.io/

      ProxyPass /trip/rest http://localhost:8080
      ProxyPassReverse /trip/rest http://localhost:8080
    </IfModule>

Redirecting to HTTPS

It is useful to ensure all users use HTTPS by providing a redirect rule to redirect any HTTP requests to use HTTPS. However, some logging clients do not support HTTP, so it may be preferable to exclude the logging patterns from redirection. Generally, the logging URLs will be of the form http://${HOST}:${PORT}/trip/rest/log_point.

This rule will redirect URLs excepting those like /trip/rest/ which can then be used by tracker clients that do not support HTTPS or redirections, to log locations without being redirected.

This rule needs to be in the <VirtualHost *:80\> section of the HTTP server.

    RedirectMatch ^/trip/app/(.*)$ https://${MY_HOST}/trip/app/$1

Configuring to redirect Traccar Client URLs

The Traccar Client app currently does not provide a facility to define a URL prefix. All calls are to the server root.

A workaround is to configure the Apache server to redirect both HTTP and HTTPS requests that match the pattern of Traccar Client logging requests to the /trip/rest/log_point URL prefix.

If you wish to support using the Traccar Client, enter the following in the Apache <VirtualHost\> sections:

    # Redirect for Traccar Client
    <IfModule mod_rewrite.c>
        RewriteEngine On
        RewriteCond "%{QUERY_STRING}" "^id=[\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12}&timestamp=\d+&lat=[-.\d]+&lon=[-.\d]+"
        RewriteRule ^/ /trip/rest/log_point [PT,QSA]
    </IfModule>

Miscellaneous

The following sections mostly relate to information around system maintenance and application development.

Tile usage monthly cumulative totals for the last year:

    SELECT year, month, max(count) AS cumulative_total FROM (
        SELECT time, extract(year from time) AS year,
        extract(month from time) AS month,
        extract(day from time) AS day,
        count FROM tile_metric ORDER BY time DESC) AS q
    GROUP BY q.year, q.month ORDER BY q.year desc, q.month DESC LIMIT 12;

Deleting expired tiles from the cache

Count how many tiles are expired:

SELECT count(*) FROM tile WHERE expires < now();

count how many are not expired:

SELECT count(*) FROM tile WHERE expires >= now();

Count how many are expired and older than 90 days:

SELECT count(*) FROM tile WHERE expires < now() AND updated < now()::timestamp::date - INTERVAL '90 days';

Delete tiles which are older than 90 days and have expired:

DELETE FROM tile WHERE expires < now() AND updated < now()::timestamp::date - INTERVAL '90 days';

Delete all expired tiles:

DELETE FROM tile WHERE expires < now();

Freeing up system disk space after deleted tiles (or other records)

To see how much space is begin used by the whole database:

SELECT pg_size_pretty(pg_database_size('trip'));

To see how much space is being used the the tiles table:

SELECT pg_size_pretty(pg_table_size('tile'));

Normally, a PostgreSQL installation will be configured to run the VACUUM command automatically from time-to-time. This allows deleted records to be re-used, but does not generally free up the system disk space being used by the deleted records. To do that, the VACUUM command needs to be run with the FULL option.

Note that VACUUM FULL requires an exclusive lock on the table it is working on so cannot be run in parallel with other database operations using the table.

See the Recovering Disk Space section of the PostgreSQL documentation for more information.

To free up the system disk space used by the tiles table, in plsql run:

VACUUM FULL tile;

or

VACUUM (FULL, VERBOSE) tile;

To free up the system disk space used by all tables:

VACUUM FULL;

or

VACUUM (FULL, VERBOSE);

Useful queries for testing

Copy location records for user with id 1 to user with id 2

    INSERT INTO location (user_id, location, "time", hdop, altitude, speed, bearing)
    SELECT 2, location, "time", hdop, altitude, speed, bearing from location where user_id = 1;

Moved yesterday's test location data forward by 1 day:

    UPDATE location SET time = time + INTERVAL '1 day'
    WHERE user_id='1' AND time >= now()::timestamp::date - INTERVAL '1 day'
    AND time <= now()::timestamp::date;

Copying data

    create table temp_location (like location);

    insert into temp_location select * from location q where user_id=29 and time >= '2015-12-14' and time <= '2015-12-14T23:59:59'

    update temp_location set user_id=3, id=nextval('location_seq'::regclass);

    insert into location select * from temp_location;

Backup

Backup just the schema, no data:

    $ pg_dump --schema-only --no-owner --no-privileges trip > schema.sql

Backup just the data, keeping the invariably large tile table separate:

    $ pg_dump --data-only --no-owner --no-privileges --exclude-table=tile trip > test-data.sql
    $ pg_dump --data-only --no-owner --no-privileges --table=tile trip > tiles.sql

Backup schema, data and privileges, including commands to recreate tables, excluding the tile data:

    $ pg_dump --clean --if-exists --no-owner --exclude-table-data=tile trip > test-schema-data.sql

The above backup is suitable for every-day backup. If you intend to restore from the backup as part of your development and test cycle, remove the tile table data exclusion so that the cache is not lost.

Deploying a Release of trip-server

See the README of the trip-web-client application for instructions on creating a release of the web client.

  1. Create backup of application's folder structure on target server

  2. Run yarn run lint

  3. Update the version number in package.json

  4. Check in the change and push the changes

  5. On the target server, pull the changes

  6. Run yarn install on the target server

  7. Optionally, run yarn outdated and compare versions with test environment

  8. If necessary run yarn upgrade

Next Release

See CHANGELOG