Client keeps trying to connect to removed account

i have logged out from and then deleted an account from my seafile desktop client (9.0.4, ubuntu 22.04). but it seems the client is unwilling to let go, as my local ~/.ccnet/logs/seafile.log is literally flooded with messaged like these:

[02/17/24 18:59:07] clone-mgr.c(702): Transition clone state for XXXXXXXX from [check server] to [error]: Failed to establish secure connection. Please check server SSL certificate.
[02/17/24 18:59:12] clone-mgr.c(686): Transition clone state for XXXXXXXX from [error] to [check server].
[02/17/24 18:59:12] clone-mgr.c(686): Transition clone state for XXXXXXXX from [error] to [check server].
[02/17/24 18:59:12] clone-mgr.c(686): Transition clone state for XXXXXXXX from [error] to [check server].
[02/17/24 18:59:12] http-tx-mgr.c(793): libcurl failed to GET https://<REMOVED SERVER>/seafhttp/protocol-version: SSL connect error.
[02/17/24 18:59:12] http-tx-mgr.c(793): libcurl failed to GET https://<REMOVED SERVER>/seafhttp/protocol-version: SSL connect error.

this goes on forever and <REMOVED SERVER> shows the URL of the server that i had completely removed from the client weeks ago. the file now has a size of 175M 210M and is growing. is there a quick fix for this?

i examined all SQLite databases in ~/Seafile/.seafile-data and found that the server that was removed was still listed in the ServerProperty table of repo.db as well as the Certs table of certs.db. but even after stopping the client, removing these rows from the tables and starting the client again, i still see above errors in seafile.log.

where does the seafile client get this outdated data from?

1 Like

Hi, you can delete the clone.db under ~/Seafile/.seafile-data, so that the client will not download libraries for logged-out accounts.

that database only seems to contain emtpy tables. i wonder where the seafile client still gets the old domain name from. that is, my problem is not that an account is simply logged out, but i removed the server completely from the client. there is nothing (visible) that still contains that server’s domain name, yet it is flooding my log files. are there any caches left to remove?

Hi, the client gets the old domain from accounts.db. But your case is caused by not clone completing this library. When you delete the clone.db and then restart the client, it should not continue to clone this library.