Package: amavisd-new Version: 1:2.11.1-5 Severity: important Hi, I recently updated my amavis/spamassassin LXC to bullseye and started experiencing random amavis shutdowns, which in turn interrupt the mail pipeline completely. That happens sometime after few minutes, sometimes after a few hours.
I've tried raising the debug level but it doesn't really help. In the logs, what I have is: Nov 9 18:57:49 lxc-amavis amavis[456]: (00456-20) Requesting process rundown after 20 tasks (and 20 sessions) Nov 9 18:57:49 lxc-amavis amavis[456]: (00456-20) TempDir removal: empty tempdir is being removed: /var/lib/amavis/tmp/amavis-20211109T115021-00456-Z7zXfwfm Nov 9 18:57:49 lxc-amavis amavis[456]: (00456-20) load: 0 %, total idle 25624.889 s, busy 22.962 s Nov 9 18:57:49 lxc-amavis amavis[456]: (00456-20) sd_notify (no socket): STOPPING=1\nSTATUS=Server rundown, notifying child processes. Nov 9 18:57:49 lxc-amavis amavis[456]: (00456-20) Net::Server: 2021/11/09-18:57:49 Server closing! Nov 9 18:57:49 lxc-amavis amavis[456]: (00456-20) sd_notify (no socket): STATUS=Child processes have been stopped. Nov 9 18:57:49 lxc-amavis amavis[456]: (00456-20) (!)Amavis::END: DB unregistering failed: The first few logs seem normal (apparently the child processes will stop after processing a number of tasks and a new child will be started) but somehow the process then fails with that DB unregistering failed message. Unfortunately, I don't have any log (with $log_level=3) after that line, so I'm unsure what actually happen. I've tried to remove the dbs in /var/lib/amavis/db in case those would be corrupted, but it didn't work, I still had the failure above. I didn't find a lot of references to that log message so I'm opening a bug here. I don't have reportbug nor net access on the LXC so I can't provide the full reportbug output but if you need information on the system please ask. It's an up to date bullseye LXC running on an up to date bullseye host. Regards, -- Yves-Alexis
