[Sorry for the top-posting; Outlook]

That all sounds helpful, but unfortunately there are no socket or service files 
(other than /usr/lib/systemd/system/[email protected], for which I have a 
clamd@scan service).  None of the RPM scripts mention anything about sockets.  
I thought it might all be down to logrotate at the end of the day, so I've 
changed the restart line to a reload line (will that close and reopen any log 
files?), but it seems something else odd is going on given that  
"/etc/logrotate.d/clamd" file contains 

create 644 root root

and all of my /var/log/clamd.scan files are owned by clamscan:clamscan...?  So 
obviously when I run "logrotate -f /etc/logrotate.d/clamd" the resulting new 
/var/log/clamd.scan is owned by root:root which obviously means clamd can't 
write to it until I manually change the ownership.

I don't understand how that's happening.  So now I have two problems (-:

Ben
--
Servers and Storage Team, UIS, University of Cambridge

-----Original Message-----
From: Kris Deugau <[email protected]> 
Sent: 02 June 2025 22:01
To: ClamAV users ML <[email protected]>
Cc: Ben Argyle <[email protected]>
Subject: Re: [clamav-users] ClamAV daemon restarts via/through systemd

Ben Argyle via clamav-users wrote:
> We're running ClamAV on RHEL 8 from EPEL 8:
> 
> clamd-1.0.8-1.el8.x86_64
> clamav-freshclam-1.0.8-1.el8.x86_64
> clamav-1.0.8-1.el8.x86_64
> 
> I know it's old, but that's what we've got to work with.  Anyway, we're 
> seeing an issue where all of our clamd@scan daemons are restarting because of 
> systemd, sometimes multiple times a day:
> 
> # journalctl -u clamd@scan | grep systemd
> [...]
> Jan 01 03:08:04 cstestapp1 systemd[1]: Stopping clamd scanner (scan) daemon...
> Jan 01 03:08:05 cstestapp1 systemd[1]: [email protected]: Succeeded.
[snip]

> It seems likely that the "01" restarts are due to logrotate:
> 
> /var/log/clamd.scan {
>      create 644 root root
>      monthly
>      compress
>      missingok
>      postrotate
>          systemctl restart clamd@scan >/dev/null 2>&1 || true
>      endscript
> }
> 
> But we can't work out what could be causing all of the other ones.  We've 
> probably ruled out OOM issues (nothing in /var/log/messages or 'dmesg').  Any 
> ideas what else I can look for, or what could be causing them?

Based on the hassles I had building a custom package for use here, I 
would guess systemd "socket activation", which is inexplicably enabled 
at least in the Debian stock packages.

clamd is very much a Very Bad use case for socket activation.

It's been a while since I piloted a RPM-based distro, but IIRC "rpm -ql 
clamd" should tell you what files got installed.  "rpm -q --scripts 
clamd" should show any pre/post-install scripts.  Look for .service and 
.sock(et?) files under (/usr)/lib/systemd/system.  (No idea which 
variation(s) might show up on RHEL.)

"systemctl edit clamd.[service|socket]" should also at least let you see 
what's configured, even if you can't find the original .service or 
.socket definitions.

-kgd
_______________________________________________

Manage your clamav-users mailing list subscription / unsubscribe:
https://lists.clamav.net/mailman/listinfo/clamav-users


Help us build a comprehensive ClamAV guide:
https://github.com/Cisco-Talos/clamav-documentation

https://docs.clamav.net/#mailing-lists-and-chat

Reply via email to