amavisd-new-2.3.2 will fail to restart when using the init script may be system speed dependant Reproducible: Always Steps to Reproduce: 1./etc/init.d/amavisd restart 2. 3. Actual Results: fails to restart Expected Results: restarted doing a stop, waiting a couple of seconds and doing a start, works fine apparently it takes few seconds to shutdown so the restart attempts to start it too early. a short pause after the stop should solve it.
Could you please try experimenting with minimal working timeout? Add "sleep x" (x being number of seconds to sleep) at the end of stop() function.
Any updates on this one?
Sorry, it kind of went on the back burner as the server hasn't been available for testing. If I can get some time weekend, I'll post the results. However, I'm not sure the stop will take the same amount of time on all systems (and this server is a dual Xeon 3GHz). Is there not a way to test for the stop before the start as opposed to picking a wait value?
I can't seem to reproduce this - `/etc/init.d/amavisd restart` works flawlessly 10 times out of 10 tries, even when in the middle of scanning of a large email. This on a P3 Celeron 667MHz.
Had it bite me again just two days ago. Made some config changes on the mail server and forgetfully just did a restart with amavis (the restart appears to work fine), some time later I noticed no email was getting delivered, examining the logs it was clear that the amavis port was offline. I then did another restart which gave me an error (the first failing restart does not provide an error (!!), only a subsequent restart attempt after that), so I did a zap, stop, then a start and all was fine. It does check for many AV products on start but we are currently on using clamd plus clamscan for backup. It has to be some sort of timing issue. Sometimes a restart will actually work but it's not dependable.
It only checks those AV products which you have uncommented in the config file.
>It only checks those AV products which you have uncommented in the config file. Yes, I'm aware of that. But I don't really mask those that are tested for by default and do enable that test for the clamav products. If anything, the extra checks should slow down the startup and ameliorate the problem.
The problem here is not the startup, though. It's the (supposedly) slow service stop. It should have nothing to do with amavisd startup.
Agreed, that's why I qualified the statement with "If anything".
Not reproducible, stale, closing.