The server received the heartbeat, but the dashboard had never updated #874
-
|
I set up a wakapi server about a year ago, and I install the wakatime extension for my firefox browser and JetBrains IDEs. Everything works at the beginning, but recently I found that the dashboard has been not updated for about 6 months: My wakapi is hosted on Zeabur with docker image setup, and I do updates for the server about two months. I use the browser almost everyday so maybe this is not issue I tried to rotate my api keys, but it seems the problem is not solved. I soon find these in my logs: It seems the server do receive the heartbeats, but for some reasons they were not updated in the dashboard. My config is stored as environment variables: How can I debug this further and solve the issue? Any help is appreciated. |
Beta Was this translation helpful? Give feedback.
Replies: 2 comments 17 replies
-
|
Very strange. Can you please check if the heartbeats actually find their way to the database? Run the following query and see if there are any recent heartbeats: select id, time, created_at from heartbeat order by created_at desc limit 100;Also check the API endpoint and see if it returns any non-zero data: Moreover, please also check for enqueued heartbeats on your client: ~/.wakatime/wakatime-cli --offline-countWhich editor plugins do you use? |
Beta Was this translation helpful? Give feedback.
-
|
I'll dig deeper into this soon! Having a busy time rn, but will get back to you sometime in the next few days, hang on. |
Beta Was this translation helpful? Give feedback.
-
|
Getting closer. I compared the heartbeats from the database you sent me with those stored on wakapi.dev. For example: id,time,hash,created_at
35695,2025-11-26 04:26:03,54050539e90b155f,2025-11-26 05:26:03.245
35723,2025-11-26 04:31:40,1c2c9021dd66305b2025-11-26 05:31:41.103In your database, these rows follow directly after each other, but with a gap in the ID range of 28. On wakapi.dev, there are 27 additional heartbeats in between those, which are missing in your database. This hints at the transaction to Can you please run you instance in Also, if you don't mind, can you try to apply the follow Git patch, recompile your Wakapi, run it and see if the issue of missing new data still persists?
Moreover, can you tell which filesystem your |
Beta Was this translation helpful? Give feedback.
-
What I can see is some I used it for a half an hour and I didn't see any rollback operations and error logs, though.
I tried the patch and build a new docker image: Before the patch: After the patch:
I asked staff from zeabur, and they say both the PostgreSQL database container and the wakapi container are on a NFS filesystem. |
Beta Was this translation helpful? Give feedback.
-
|
The
Ah, sorry, I assumed you were using SQLite as a database, which doesn't work reliable over network filesystems. With Postgres, everything should be fine, though. Judging from the latest excerpt form your heartbeats table, things look pretty much alright, don't they? At least I don't see those large gaps between heartbeats anymore? But the problem still persists? |
Beta Was this translation helpful? Give feedback.
-
No, I think the major problem is solved now.
Yes, I think so. These things seem working again now. But I'm still trying to sort out one thing: according to the quoted part, those small id gaps between the heartbeats are normal, aren't they? |
Beta Was this translation helpful? Give feedback.
-
Yes, sort of. There are essentially two potential causes for "missing" IDs. Either the heartbeats were actually not inserted to to some error (i.e. the database transaction was rolled back) or they were exact duplicates (same timestamp, same file, etc.) of an already existing heartbeat. Even though I still don't have much of an idea what had caused your initial problem, I'll close this discussion for now, as things seem to be back to normal again. Perhaps there was some issue with the database connection at your hosting provider. Feel free to get back here if you're experiencing a similar issue again. Regarding #877: will investigate soon. |
Beta Was this translation helpful? Give feedback.
Yes, sort of. There are essentially two potential causes for "missing" IDs. Either the heartbeats were actually not inserted to to some error (i.e. the database transaction was rolled back) or they were exact duplicates (same timestamp, same file, etc.) of an already existing heartbeat.
Even though I still don't have much of an idea what had caused your initial problem, I'll close this discussion for now, as things seem to be back to normal again. Perhaps there was some issue with the database connection at your hosting provider. Feel free to get back here if you're experiencing a similar issue again.
Regarding #877…