#79 Start script should try to get pid file from opendkim.conf
opendkim -V
opendkim: OpenDKIM Filter v2.4.0
Compiled with OpenSSL 1.0.0d 8 Feb 2011
SMFI_VERSION 0x1000001
libmilter version 1.0.1
Supported signing algorithms:
rsa-sha1
rsa-sha256
Supported canonicalization algorithms:
relaxed
simple
Active code options:
USE_DB
libopendkim 2.4.0:
ldd `which opendkim`
linux-gate.so.1 => (0xf57fe000)
libopendkim.so.4 => /usr/local/lib/libopendkim.so.4 (0xb7714000)
libresolv.so.2 => /lib/libresolv.so.2 (0xb76fd000)
libssl.so.1.0.0 => /usr/local/ssl/lib/libssl.so.1.0.0 (0xb76ab000)
libcrypto.so.1.0.0 => /usr/local/ssl/lib/libcrypto.so.1.0.0 (0xb752f000)
libdl.so.2 => /lib/libdl.so.2 (0xb752a000)
libz.so.1 => /lib/libz.so.1 (0xb7513000)
libdb-4.5.so => /usr/lib/libdb-4.5.so (0xb73ce000)
libpthread.so.0 => /lib/libpthread.so.0 (0xb73b3000)
libc.so.6 => /lib/libc.so.6 (0xb7246000)
/lib/ld-linux.so.2 (0xb774b000)
opendkim -x /usr/local/etc/opendkim.conf
opendkim: smfi_opensocket() failed
Not sure what other/additional info is needed or helpful.
The contents of your opendkim.conf would be helpful.
Is this a new install or an upgrade? Which distro and version? Seems like you compiled from source since you are using /usr/local/etc/opendkim.conf.
egrep -i ^Socket /usr/local/etc/opendkim.conf
If it's a local unix socket, make sure the path can be accessed by the user that opendkim runs as. If it's a TCP socket, make sure nothing is listening already on that port:
CentOS49[root@smtp4 ~]# egrep -i ^socket /usr/local/etc/opendkim.conf
Socket inet:49999@localhost
CentOS49[root@smtp4 ~]# netstat -ltnp | grep :49999
tcp 0 0 127.0.0.1:49999 0.0.0.0:* LISTEN 5892/opendkim
it looks like the PID's not acting like I think it should.
> Is this a new install or an upgrade?
new install. well, technically, a new build installed over an old install.
> Which distro and version?
opensuse 11.4/64
> Seems like you compiled from source since you are using /usr/local/etc/opendkim.conf.
that's correct.
> egrep -i ^Socket /usr/local/etc/opendkim.conf
egrep -i ^Socket /usr/local/etc/opendkim.conf
Socket inet:9999@localhost
> If it's a TCP socket, make sure nothing is listening already on that port:
service opendkim start
opendkim: smfi_opensocket() failed
netstat -ltnp | grep :9999
tcp 0 0 127.0.0.1:9999 0.0.0.0:* LISTEN 12877/opendkim
service opendkim stop
OpenDkim is already stopped.
netstat -ltnp | grep :9999
tcp 0 0 127.0.0.1:9999 0.0.0.0:* LISTEN 12877/opendkim
???
kill -9 12877
netstat -ltnp | grep :9999
tcp 0 0 127.0.0.1:9999 0.0.0.0:* LISTEN 12877/opendkim
really ???
shutdown -r now
tic ... toc ... tic ... toc ...
netstat -ltnp | grep :9999
service opendkim-custom start
Starting OpenDkim.
NO Error ...
netstat -ltnp | grep :9999
tcp 0 0 127.0.0.1:9999 0.0.0.0:* LISTEN 11927/opendkim
opendkim -x /usr/local/etc/opendkim.conf
opendkim: smfi_opensocket() failed
service opendkim stop
OpenDkim is already stopped.
netstat -ltnp | grep :9999
tcp 0 0 127.0.0.1:9999 0.0.0.0:* LISTEN 12927/opendkim
Confusion.
> The contents of your opendkim.conf would be helpful.
cat /etc/init.d/opendkim
----------------
#!/bin/sh
set -e
### BEGIN INIT INFO
# Provides: opendkim
# Required-Start: $local_fs $remote_fs $network $syslog $time
# Required-Stop: $local_fs $remote_fs $network $syslog
# Should-Start:
# Should-Stop:
# Default-Start: 3 5
# Default-Stop: 0 1 2 4 6
# Short-Description: OpenDkim init script
# Description: Init script for OpenDkim services
### END INIT INFO
OPENDKIM_BIN=/usr/local/sbin/opendkim
OPENDKIM_CONF=/usr/local/etc/opendkim.conf
OPENDKIM_PID=/var/run/opendkim.pid
# Uncomment to allow Dovecot daemons to produce core dumps.
#ulimit -c unlimited
if test -f $OPENDKIM_PID; then
running=yes
else
running=no
fi
case "$1" in
start)
echo -n "Starting OpenDkim"
$OPENDKIM_BIN -x $OPENDKIM_CONF
echo "."
;;
stop)
if test $running = yes; then
echo "Stopping OpenDkim"
kill `cat $OPENDKIM_PID`
echo "."
else
echo "OpenDkim is already stopped."
fi
;;
restart)
$0 stop
$0 start
echo "."
;;
*)
echo "Usage: /etc/init.d/opendkim-custom {start|stop|restart}" >&2
exit 1
;;
esac
exit 0
----------------
killall opendkim
service opendkim start
Starting OpenDkim.
cat /var/run/opendkim.pid
cat: /var/run/opendkim.pid: No such file or directory
???
grep -i pid /usr/local/etc/opendkim.conf
PidFile /var/run/opendkim.pid
cat /usr/local/etc/opendkim.conf
----------------
Background Yes
Canonicalization relaxed/simple
ClockDrift 300
Diagnostics yes
DNSTimeout 10
Domain eval.loc
DomainKeysCompat no
EnableCoredumps no
FixCRLF no
InternalHosts /var/db/dkim/internal_hosts
KeepTemporaryFiles no
KeyFile /home/eval/dkim/dkim.eval.loc.pem
LogWhy yes
MaximumHeaders 65536
MilterDebug 0
Minimum 0
Mode sv
MultipleSignatures no
NoHeaderB no
PidFile /var/run/opendkim.pid
Quarantine no
ReportAddress eval@localhost
RequireSafeKeys Yes
ResolverTracing no
Selector key1.eval
SendADSPReports No
SendReports No
SignatureAlgorithm rsa-sha256
SignatureTTL 0
Socket inet:9999@localhost
StrictHeaders no
StrictTestMode no
SubDomains No
Syslog yes
SyslogFacility mail
SyslogSuccess yes
TemporaryDirectory /var/tmp
UserID dkim:dkim
X-Header yes
----------------
opendkim -x /usr/local/etc/opendkim.conf -P /var/run/opendkim.pid
ls -al /var/run/opendkim.pid
/bin/ls: cannot access /var/run/opendkim.pid: No such file or directory
killall opendkim
mkdir -p /var/run/opendkim
chown dkim:dkim /var/run/opendkim
opendkim -x /usr/local/etc/opendkim.conf -P /var/run/opendkim/opendkim.pid
ls -al /var/run/opendkim/opendkim.pid
-rw-r--r-- 1 dkim dkim 6 Jun 6 14:27 /var/run/opendkim/opendkim.pid
cat /var/run/opendkim/opendkim.pid
13113
ps ax | grep opendkim
13113 ? Ssl 0:00 opendkim -x /usr/local/etc/opendkim.conf -P /var/run/opendkim/opendkim.pid
Changing to THAT pid dir in the init.d should do it.
If you're using contrib/init/redhat/opendkim as your start/stop script, it just means your config file and the pid file it expects aren't the same.
You can have the auto-build stuff use a matching pathname by specifying "--localstatedir=/var" to ./configure.
> If you're using contrib/init/redhat/opendkim as your start/stop script, it
> just means your config file and the pid file it expects aren't the same.
i of course mad any mods necessary before launch -- this was, after all, running find with opendkim 2.3.2.
both @ cmd line, and using PidFile spec in config file should be sufficient to override any/all build-time localstatedir spec, right?
the issue seems to have been that, since exec was as uid:gid(dkim:dkim), that perms on /var/run were no longer sufficient. creating /var/run/opendkim, setting it to dkim:dkim ownership, and putting the pid in there did the trick.
in any case, all better now.
Changing the description accordingly. Will try to fix this for 2.4.1.
Done for 2.4.1.
This change to the script has been reverted. After talking to other Linux users in our community, consensus appears to be that if you are going to use a pid file different from the one hard-coded in the script, you're tinkering enough already that you should be able to make your config file and the script match on your own.
Nevertheless, 2.4.1 includes a method to extract config values via the binary (new "-e" switch) so you can hack this back in easily if you like.
Fix for this bug included in new release.