systemctl versus service

[Robert Moskowitz]

On 04/17/2014 02:20 PM, poma wrote:
> On 17.04.2014 20:07, Robert Moskowitz wrote:
>> So for a number of versions, 'service' has been a front end to
>> 'systemctl'. I have a service I am testing the Host Identity Protocol,
>> and the HIPL implementation: http://infrahip.hiit.fi/ that is still
>> built around 'service'. And they maintain binaries for a number of
>> distros, including Centos that do not yet (if ever?) use 'systemctl'.
>>
>> Now I amy not have all the terms right, but that is pretty close to the
>> challenge.
>>
>> On my Fedora 20 system, the services come up real slow. In fact I get
>> errors that they failed, even though they are working! Or at least they
>> are testing out properly. For example, when I start the HIP firewall I get:
>>
>> # service hipfw start
>> Starting hipfw (via systemctl): Job for hipfw.service failed. See
>> 'systemctl status hipfw.service' and 'journalctl -xn' for details.
>> [FAILED]
>>
>> This takes a LONG time until the failed message appears. 'ps axu|grep
>> hip' shows the service is running. but status shows:
>>
>> hipfw.service - SYSV: HIPL firewall daemon
>> Loaded: loaded (/etc/rc.d/init.d/hipfw)
>> Active: failed (Result: timeout) since Thu 2014-04-17 12:51:05 EDT; 1h
>> 0min ago
>> Process: 2029 ExecStart=/etc/rc.d/init.d/hipfw start (code=exited,
>> status=0/SUCCESS)
>> CGroup: /system.slice/hipfw.service
>> └─2031 /usr/sbin/hipfw -bklpF
>>
>> Apr 17 12:46:05 ee900.htt-consult.com hipfw[2031]:
>> debug(hipfw/helpers.c:97 at system_print) $ ip6tables -I HIPFW-FORWARD -p
>> 139 -j DROP -> 0
>> Apr 17 12:46:05 ee900.htt-consult.com hipfw[2031]:
>> debug(hipfw/helpers.c:97 at system_print) $ iptables -I INPUT -j
>> HIPFW-INPUT -> 0
>> Apr 17 12:46:05 ee900.htt-consult.com hipfw[2031]:
>> debug(hipfw/helpers.c:97 at system_print) $ iptables -I OUTPUT -j
>> HIPFW-OUTPUT -> 0
>> Apr 17 12:46:05 ee900.htt-consult.com hipfw[2031]:
>> debug(hipfw/helpers.c:97 at system_print) $ iptables -I FORWARD -j
>> HIPFW-FORWARD -> 0
>> Apr 17 12:46:05 ee900.htt-consult.com hipfw[2031]:
>> debug(hipfw/helpers.c:97 at system_print) $ ip6tables -I INPUT -j
>> HIPFW-INPUT -> 0
>> Apr 17 12:46:05 ee900.htt-consult.com hipfw[2031]:
>> debug(hipfw/helpers.c:97 at system_print) $ ip6tables -I OUTPUT -j
>> HIPFW-OUTPUT -> 0
>> Apr 17 12:46:05 ee900.htt-consult.com hipfw[2031]:
>> debug(hipfw/helpers.c:97 at system_print) $ ip6tables -I FORWARD -j
>> HIPFW-FORWARD -> 0
>> Apr 17 12:46:05 ee900.htt-consult.com hipfw[2031]:
>> debug(libcore/message.c:319 at sendto_hipd) Sending user message 16 to HIPD
>> on socket 4
>> Apr 17 12:46:05 ee900.htt-consult.com hipfw[2031]:
>> debug(libcore/message.c:323 at sendto_hipd) Sent 40 bytes
>> Apr 17 12:46:05 ee900.htt-consult.com hipfw[2031]:
>> debug(libcore/message.c:400 at send_recv_info_internal) Waiting to receive
>> daemon info.
>> Apr 17 12:51:05 ee900.htt-consult.com systemd[1]: hipfw.service
>> operation timed out. Terminating.
>> Apr 17 12:51:05 ee900.htt-consult.com systemd[1]: Failed to start SYSV:
>> HIPL firewall daemon.
>> Apr 17 12:51:05 ee900.htt-consult.com systemd[1]: Unit hipfw.service
>> entered failed state.
>>
>> Well the HIPL developers are looking into supporting systemctl method,
>> but for the present, I need the start not to take so long (and I am
>> talking minutes here, not a few extra seconds until the failed message).
>>
>> So is there some option I could use in 'systemctl start hipfw.service'
>> that will cut down the start time, even if there are some warnings?
>>
>> thanks
>>
>>
> Host Identity Protocol for Linux - Systemd
> https://bugs.launchpad.net/hipl/+bug/1302949
I know about the bug report.  Miika entered it after I reported the problem.

> Seek advice at
> systemd-devel -- systemd Development Mailing List
> http://lists.freedesktop.org/mailman/listinfo/systemd-devel

I will make sure the developers have this reference.  If they don't 
already.  It is all a matter of time...

thanks