Constant I/O load

My LXC (version 3.15) container produces a constant I/O load, i.e., my disk is constantly being accessed. The CPU load on the other hand is not particularly high. On the host, I observed via iotop that [jdbX/dm-X-X], [txg_sync] and lxd have a high relative I/O activity. Also, I noticed that the lxd process is being “restarted” regularly while the container is running, i.e., the process number keeps changing in intervals of a few minutes or so. Is that normal?
By “lxd process” I mean the process spawned from “/bin/sh /snap/lxd/11270/commands/daemon.start” inside the container. In fact, the process “/bin/sh /snap/lxd/11270/commands/daemon.start” and all its child processes keep changing their pid.

In the container I noted that there is a contant low CPU activity of snapfuse (core).

Setting zfs_prefetch_disable to 1 does not solve the problem.

Can you tell me potential reasons for this behavior and how to investigate it further? Thank you.

smartctl tells me that my disk has no issues.