Query database

Pi-hole FTLDNS uses the well-known relational database management system SQLite3 as its long-term storage of query data. In contrast to many other database management solutions, FTLDNS does not need a server database engine as the database engine is directly embedded in FTLDNS. It seems an obvious choice as it is probably the most widely deployed database engine - it is used today by several widespread web browsers, operating systems, and embedded systems (such as mobile phones), among others. Hence, it is rich in supported platforms and offered features. SQLite implements most of the SQL-92 standard for SQL and can be used for high-level queries.

The long-term query database was the first database that was added to the Pi-hole project. We update this database periodically and on the exit of FTLDNS (triggered e.g. by a service pihole-FTL restart). The updating frequency can be controlled by the parameter DBINTERVAL and defaults to once per minute. We think this interval is sufficient to protect against data losses due to power failure events. FTLDNS needs the database to populate its internal history of the most recent 24 hours. If the database is disabled, FTLDNS will show an empty query history after a restart.

The location of the database can be configured by the config parameter DBFILE. It defaults to /etc/pihole/pihole-FTL.db. If the given file does not exist, FTLDNS will create a new (empty) database file.

Another way of controlling the size of the long-term database is by setting a maximum age for log queries to keep using the config parameter MAXDBDAYS. It defaults to 365 days, i.e. queries that are older than one year get periodically removed to limit the growth of the long-term database file.

The config parameter DBIMPORT controls whether FTL loads information from the database on startup. It needs to do this to populate the internal data structure with the most recent history. However, as importing from the database on disk can delay FTL on very large deploys, it can be disabled using this option.

Split database¶

You can split your long-term database by periodically rotating the database file (do this only when pihole-FTL is not running). The individual database contents can easily be merged when required. This could be implemented by running a monthly cron job such as:

sudo service pihole-FTL stop
sudo mv /etc/pihole/pihole-FTL.db /media/backup/pihole-FTL_$(date +"%m-%y").db
sudo service pihole-FTL start

Note that DNS resolution will not be available as long as pihole-FTL is stopped.

Backup database¶

The database can be backed up while FTL is running when using the SQLite3 Online backup method, e.g.,

sqlite3 /etc/pihole/pihole-FTL.db ".backup /home/pi/pihole-FTL.db.backup"

will create /home/pi/pihole-FTL.db.backup which is a copy of your long-term database.

The long-term database contains several tables:

Query Table¶

Label	Type	Allowed to by empty	Content
`id`	integer	No	autoincrement ID for the table, only used by SQLite3, not by FTLDNS
`timestamp`	integer	No	Unix timestamp when this query arrived at FTLDNS (used as index)
`type`	integer	No	Type of this query (see Supported query types)
`status`	integer	No	How was this query handled by FTLDNS? (see Supported status types)
`domain`	text	No	Requested domain
`client`	text	No	Requesting client (IP address)
`forward`	text	Yes	Forward destination used for this query (only set if `status == 2`)
`additional_info`	blob	Yes	Data-dependent content, see below
`reply_type`	integer	Yes	Type of the reply for this query (see Supported reply types)
`reply_time`	real	Yes	Seconds it took until the final reply was received
`dnssec`	integer	Yes	Type of the DNSSEC status for this query (see DNSSEC status)
`regex_id`	integer	Yes	ID of the regex filter that matched this query (only set if blocked by a regex filter)

The queries VIEW is dynamically generated from the data actually stored in the query_storage table and the linking tables domain_by_id, client_by_id, forward_by_id, and addinfo_by_id (see below). The table query_storage will contains integer IDs pointing to the respective entries of the linking tables to save space and make searching the database faster. If you haven't upgraded for some time, the table may still contain strings instead of integer IDs.

Data-dependent `additional_info` field¶

The content and type of the additional_info row depends on the status of the given query. For many queries, this field is empty. You should, however, not rely on this field being empty as we may add content of any type for other status types in future releases.

Query blocked due to a CNAME inspection (status 9, 10, 11)¶

If a query was blocked due to a CNAME inspection (status 9, 10, 11), this field contains the domain which was the reason for blocking the entire CNAME chain (text).

Query influenced by a black- or whitelist entry¶

If a query was influenced by a black- or whitelist entry, this field contains the ID of the corresponding entry in the domainlist table.

Counters table¶

This table contains counter values integrated over the entire lifetime of the table

Label	Type	Allowed to by empty	Content
`id`	integer	No	ID for the table used to select a counter (see below)
`value`	integer	No	Value of a given counter

ID	Interpretation
0	Total number of queries
1	Total number of blocked queries

FTL table¶

The FTL table contains some data used by FTLDNS for determining which queries to save to the database. This table does not contain any entries of general interest.

Supported query types¶

ID	Resource Record (a.k.a. query type)
1	`A`
2	`AAAA`
3	`ANY`
4	`SRV`
5	`SOA`
6	`PTR`
7	`TXT`
8	`NAPTR`
9	`MX`
10	`DS`
11	`RRSIG`
12	`DNSKEY`
13	`NS`
14	`OTHER` (any query type not covered elsewhere, but see note below)
15	`SVCB`
16	`HTTPS`

Any other query type will be stored with an offset of 100, i.e., TYPE66 will be stored as 166 in the database (see pi-hole/FTL #1013). This is done to allow for future extensions of the query type list without having to change the database schema. The OTHER query type is deprecated since Pi-hole FTL v5.4 (released Jan 2021) and not used anymore. It is kept for backwards compatibility. Note that OTHER is still used for the regex extension querytype= filter and used for all queries not covered by the above list.

Supported status types¶

ID	Status		Details
0	Unknown	❔	Unknown status (not yet known)
1	Blocked	❌	Domain contained in gravity database
2	Allowed	✅	Forwarded
3	Allowed	✅	Replied from cache
4	Blocked	❌	Domain matched by a regex blacklist filter
5	Blocked	❌	Domain contained in exact blacklist
6	Blocked	❌	By upstream server (known blocking page IP address)
7	Blocked	❌	By upstream server (`0.0.0.0` or `::`)
8	Blocked	❌	By upstream server (`NXDOMAIN` with `RA` bit unset)
9	Blocked	❌	Domain contained in gravity database Blocked during deep CNAME inspection
10	Blocked	❌	Domain matched by a regex blacklist filter Blocked during deep CNAME inspection
11	Blocked	❌	Domain contained in exact blacklist Blocked during deep CNAME inspection
12	Allowed	✅	Retried query
13	Allowed	✅	Retried but ignored query (this may happen during ongoing DNSSEC validation)
14	Allowed	✅	Already forwarded, not forwarding again
15	Blocked	❌	Blocked (database is busy) How these queries are handled can be configured
16	Blocked	❌	Blocked (special domain) E.g. Mozilla's canary domain and Apple's Private Relay domains Handling can be configured
17	Allowed	✅⌛	Replied from stale cache

Supported reply types¶

ID	Reply type is
0	unknown (no reply so far)
1	`NODATA`
2	`NXDOMAIN`
3	`CNAME`
4	a valid `IP` record
5	`DOMAIN`
6	`RRNAME`
7	`SERVFAIL`
8	`REFUSED`
9	`NOTIMP`
10	`OTHER`
11	`DNSSEC`
12	`NONE` (query was dropped intentionally)
13	`BLOB` (binary data)

DNSSEC status¶

ID	DNSSEC status is
0	unknown
1	`SECURE`
2	`INSECURE`
3	`BOGUS`
4	`ABANDONED`
5	`TRUNCATED`

Linking tables¶

The queries VIEW reads repeating properties from linked tables to reduce both database size and search complexity. These linking tables, domain_by_id, client_by_id, forward_by_id, and addinfo_by_id all have a similar structure:

`domain_by_id`¶

Label	Type	Allowed to by empty	Content
`id`	integer	No	ID of the entry. Used by `query_storage`
`domain`	text	No	Domain name

`client_by_id`¶

Label	Type	Allowed to by empty	Content
`id`	integer	No	ID of the entry. Used by `query_storage`
`ip`	text	No	Client IP address
`name`	text	Yes	Client host name

`forward_by_id`¶

Label	Type	Allowed to by empty	Content
`id`	integer	No	ID of the entry. Used by `query_storage`
`forward`	text	No	Upstream server identifier (`<ipaddr>#<port>`)

`addinfo_by_id`¶

Label	Type	Allowed to by empty	Content
`id`	integer	No	ID of the entry. Used by `query_storage`
`type`	integer	No	Type of the `content` field
`content`	blob	No	Type-dependent content

Valid type IDs are currently

ADDINFO_CNAME_DOMAIN = 1 - content is a string (the related CNAME)
ADDINFO_DOMAIN_ID = 2 - content is an integer (ID pointing to a domain in the domainlist table)

Example for interaction with the long-term query database¶

In addition to the interactions the Pi-hole database API offers, you can also run your own SQL commands against the database. If you want to obtain the three most queries domains for all time, you could use

sqlite3 "/etc/pihole/pihole-FTL.db" "SELECT domain,count(domain) FROM queries WHERE (STATUS == 2 OR STATUS == 3) GROUP BY domain ORDER BY count(domain) DESC LIMIT 3"

which would return something like

discourse.pi-hole.net|421095
www.pi-hole.net|132483
posteo.de|130243

showing the domain and the number of times it was found in the long-term database. Note that such a request might take a very long time for computation as the entire history of queries has to be processed for this.