Unable to start teslamate after upgrading from v2.0.0 to v2.1.0

[han9over]

Posting this here because not getting any help in the issue section. Request got closed without any information. Title got changed that I messed up with DB which is incorrect. #4868

When upgrading to 2.0.0, I followed the upgrade instructions provided. Took backup. Upgraded Postgres to 17. Then updated Teslamate to 2.0.0. Then restored from backup. This was done couple days after 2.0.0 is released. No issues after it and everything worked.

After 2.1.0 released, I did normal update process of backup, pull, up -d. then teslamate didn’t start and found error in logs. Below is the error message log.

2025-08-02T18:26:57.920149679Z 2025-08-02 14:26:57.913 [info] == Running 20250613133700 TeslaMate.Repo.Migrations.AddAndCalculateElevationChanges.up/0 forward
2025-08-02T18:26:57.921293787Z 2025-08-02 14:26:57.919 [info] alter table drives
2025-08-02T18:26:57.929180408Z 2025-08-02 14:26:57.928 [info] execute "WITH elevation_changes AS (\n  SELECT\n    drive_id,\n    SUM(CASE WHEN elevation_diff > 0 THEN elevation_diff ELSE 0 END) as ascent,\n    SUM(CASE WHEN elevation_diff < 0 THEN ABS(elevation_diff) ELSE 0 END) as descent\n  FROM (\n    SELECT\n      drive_id,\n      elevation - LAG(elevation) OVER (PARTITION BY drive_id ORDER BY date) as elevation_diff\n    FROM positions\n    WHERE elevation IS NOT NULL\n  ) as changes\n  GROUP BY drive_id\n)\nUPDATE drives d\nSET\n  ascent = ec.ascent,\n  descent = ec.descent\nFROM elevation_changes ec\nWHERE d.id = ec.drive_id;\n"
2025-08-02T18:27:10.808122430Z ** (Postgrex.Error) ERROR 22003 (numeric_value_out_of_range) smallint out of range
2025-08-02T18:27:10.808222134Z     (ecto_sql 3.12.1) lib/ecto/adapters/sql.ex:1096: Ecto.Adapters.SQL.raise_sql_call_error/1
2025-08-02T18:27:10.808232486Z     (elixir 1.17.3) lib/enum.ex:1703: Enum."-map/2-lists^map/1-1-"/2
2025-08-02T18:27:10.808239856Z     (ecto_sql 3.12.1) lib/ecto/adapters/sql.ex:1203: Ecto.Adapters.SQL.execute_ddl/4
2025-08-02T18:27:10.808246615Z     (ecto_sql 3.12.1) lib/ecto/migration/runner.ex:348: Ecto.Migration.Runner.log_and_execute_ddl/3
2025-08-02T18:27:10.808253412Z     (elixir 1.17.3) lib/enum.ex:1703: Enum."-map/2-lists^map/1-1-"/2
2025-08-02T18:27:10.808260023Z     (elixir 1.17.3) lib/enum.ex:1703: Enum."-map/2-lists^map/1-1-"/2
2025-08-02T18:27:10.808266726Z     (ecto_sql 3.12.1) lib/ecto/migration/runner.ex:311: Ecto.Migration.Runner.perform_operation/3

[brianmay]

Wonder if your elevation data is bad (elevation is a float), which is causing the ascent/descent values to overflow (smallint range -32,768 to +32,767- should be more then enough).

I believe elevation is in meters (need to check this).

e.g. if there is a jump from elevation from 0 to 33,000 meters between to adjacent positions. But unless you have your car on a rocket or something, that would seem unlikely.

You just need one bad reading to cause a crash like this. Finding the bad reading may not be so easy, depending on your SQL skill level and/or ability to find something that looks off in the grafana dashboards.

[han9over]

Wonder if your elevation data is bad (elevation is a float), which is causing the ascent/descent values to overflow (smallint range -32,768 to +32,767- should be more then enough).

I believe elevation is in meters (need to check this).

e.g. if there is a jump from elevation from 0 to 33,000 meters between to adjacent positions. But unless you have your car on a rocket or something, that would seem unlikely.

You just need one bad reading to cause a crash like this. Finding the bad reading may not be so easy, depending on your SQL skill level and/or ability to find something that looks off in the grafana dashboards.

i can handle Sql. Let me try tomorrow and get back.

[han9over]

I was able to find the bad drive which had incorrect elevation. fixed it. Now upgraded to 2.1.0 and back on the latest. Thanks for the help.

[han9over]

If anyone looking for the solution, I used HeidiSQL tool to connect to the database.

/* Run this to make sure your positions query isn't timing out */
SET statement_timeout = 0;

/* run this to find the bad elevation data for drives */
WITH elevation_changes AS (
SELECT
drive_id,
SUM(CASE WHEN elevation_diff > 0 THEN elevation_diff ELSE 0 END) AS ascent,
SUM(CASE WHEN elevation_diff < 0 THEN ABS(elevation_diff) ELSE 0 END) AS descent
FROM (
SELECT
drive_id,
elevation - LAG(elevation) OVER (PARTITION BY drive_id ORDER BY date) AS elevation_diff
FROM positions
WHERE elevation IS NOT NULL
) AS changes
GROUP BY drive_id
)
SELECT *
FROM elevation_changes
WHERE ascent > 32767 OR descent > 32767;

/* You can run this to get details of the drive and view the drive in grafana to find the elevation graph */
SELECT * FROM positions WHERE drive_id = <>;

/* You can either clear the elevation data for that drive or set anything above 300 to something lower */
UPDATE positions SET elevation = <> WHERE drive_id = <> AND elevation > <<limit it to less than 100 to be safe>>;