1 |
On Sat, Feb 22, 2014 at 6:16 PM, <thegeezer@×××××××××.net> wrote: |
2 |
> OK so because of how much time has been spent arguing about systemd with |
3 |
> little technical content, i've spent some time on the freedesktop site |
4 |
> reading Lennart's blog and also going through the source to find answers |
5 |
> to my questions about the socket activator. |
6 |
> i've also been going through the man pages of netctl too and am horrified |
7 |
> at the lack of what i would call enterprise features. |
8 |
|
9 |
networkd (netctl is just the command-line front-end) is not intended |
10 |
for enterprise; it's for little servers where you only need static IPs |
11 |
or simple bridges. For desktops/laptops, you are supposed to keep |
12 |
using NetworkManager/connman/whatever you used before. For complex |
13 |
network setups, you need *a* network manager (not necessarily |
14 |
NetworkManager). |
15 |
|
16 |
> this is by no means a definitive list. |
17 |
> I just thought that i would share what i had found. |
18 |
> please correct me if i am wrong in any of these. |
19 |
> please add to the list for technical items only. |
20 |
|
21 |
I find it a very impartial and objective review; thank you very much! |
22 |
|
23 |
> thanks! |
24 |
> |
25 |
> pros |
26 |
> 1.very modular, everything can be disabled though not removed |
27 |
> 2.socket based activator allows restart of services with no service |
28 |
> interruption |
29 |
> 3.if activator.c is used for this, then the code is actually pretty clean |
30 |
> using supplied sd-daemon.c simplifies sockets for daemons and also adds |
31 |
> extra watchdog features |
32 |
> 4.can disable socket based activation according to Canek, but i can't find |
33 |
> how. |
34 |
|
35 |
You use a .service unit file instead of a .socket unit file. That's it. |
36 |
|
37 |
For OpenSSH, for example, you can enable sshd.service[1], and then the |
38 |
SSH daemon works as it does in OpenRC. If you instead enable |
39 |
sshd.socket[2], then the daemon will start on demand. |
40 |
|
41 |
You don't have to *disable* anything; you choose how do you want to |
42 |
use your services (if the services provide both ways, like OpenSSH |
43 |
does). |
44 |
|
45 |
> 5.fschecking mounts and logging output (though how for corrupt / notsure) |
46 |
|
47 |
Corrupt filesystems or logs? |
48 |
|
49 |
> 6.auto-gettys allows for lower numbered X windows by default for e.g. |
50 |
> multiseat and dynamic serial ttys |
51 |
> 7.clever logging, including from nspawned containers' logs and distributed |
52 |
> for enterprise |
53 |
> 8.nspawning using filename namespaces |
54 |
> 9.systemctl kill <service> -- killing service and all forks and spawn |
55 |
> cgtop -- top with cgroups |
56 |
> 10.much easier to define resource limitations per service |
57 |
> |
58 |
> cons |
59 |
> 1.new tools to learn, new gotchas to learn. |
60 |
> 2.yet to go through systemd source to find out how modular or not it is. |
61 |
|
62 |
While it tries to be modular where it can, systemd prefers simple code |
63 |
and integrated solutions. Modularity is not going to be one of its |
64 |
strong points. |
65 |
|
66 |
> 3.not clear how the socket activator works, the code activator.c appears |
67 |
> to be to _test_ activation only, with activator code being elsewhere. if |
68 |
> it is used then you would have one process running for each port it is |
69 |
> virtually listened to. |
70 |
|
71 |
It's been a while since I've read the source code, but it isn't in |
72 |
src/activate/activate.c[3]? |
73 |
|
74 |
> 4./etc/machine-id because hostname and node id in the <cluster of your |
75 |
> choice> are not enough. |
76 |
|
77 |
The idea is that machine-id is as unique as reasonable to ask. I'm not |
78 |
overly happy with it, too, but that's the justification. |
79 |
|
80 |
Imagine thousands of virtual machines running services, and you want |
81 |
to coalesce all their journal logs in a central server. With |
82 |
machine-id, you don't need to worry even to change the default |
83 |
"localhost" for your throwaway VMs, you can detect the different logs |
84 |
immediately (machine-id should be generated at OS install time; for |
85 |
rolling distros, I think they generate it if when installing systemd |
86 |
is not available.) |
87 |
|
88 |
> 5./fsck.options gives more options than "auto""force""skip" on reboot |
89 |
> 6.requiring logging tools in rescue cds in order to view logs |
90 |
|
91 |
Yeah, that's a drag. However, you *can* run rsyslog (or syslog-ng) |
92 |
alongside the journal, and have the best of both worlds. Or you can |
93 |
automatically send the journal logs to a central server designed for |
94 |
that purpose only. |
95 |
|
96 |
> 7.chroots no longer work. forcing use of nspawn to ensure environment set |
97 |
> up correctly. |
98 |
|
99 |
I'm sorry, chroot doesn't work? First time I heard about it. While |
100 |
systemd-nspawn is a gazillion times better than a simple chroot, you |
101 |
*can* still use a chroot if you so desire. Where did you found that |
102 |
chroot doesn't works? |
103 |
|
104 |
> 8.strange gotchas: that because of socket/dbus etc activation you have to |
105 |
> disable a service first, then stop it in case it is then restarted in the |
106 |
> background |
107 |
|
108 |
You only need to stop the socket too. You can do it at the same time: |
109 |
|
110 |
systemctl stop avahi-daemon.service avahi-daemon.socket |
111 |
|
112 |
> 9.the new deal breaker for me is the networking. |
113 |
> for anything remotely complex (i.e. two IP addresses on an interface woo), |
114 |
> need to use netctl. |
115 |
> a.which doesn't support vlan naming types i.e. padding zeroes |
116 |
> b.doesn't appear to support gre keys |
117 |
> c.doesn't appear to support multiple routing tables |
118 |
> d.doesn't appear to support "ip rule" |
119 |
> e.doesn't have lacp support for bonding |
120 |
> f.there is the option for running a script in PRE and POST UP but...no |
121 |
|
122 |
networkd (again, netctl is the command-line front-end) is not for |
123 |
enterprise networks; on the contrary, is for the trivial cases. For |
124 |
example, in a little web server I administer I have: |
125 |
|
126 |
$ cat /etc/systemd/system/network.service |
127 |
[Unit] |
128 |
Description=Static network service |
129 |
After=local-fs.target |
130 |
Before=network.target |
131 |
Documentation=man:ifconfig(8) |
132 |
Documentation=man:route(8) |
133 |
|
134 |
[Service] |
135 |
Type=oneshot |
136 |
RemainAfterExit=yes |
137 |
ExecStart=/bin/ifconfig enp2s12 192.168.1.2 broadcast 192.168.1.255 |
138 |
netmask 255.255.255.0 up |
139 |
ExecStart=/bin/route add default gw 192.168.1.1 enp2s12 |
140 |
|
141 |
[Install] |
142 |
WantedBy=multi-user.target |
143 |
|
144 |
(Yeah, I know, I should switch to ip, I'm sorry, I haven't had the time yet). |
145 |
|
146 |
I'm going to get rid of this trivial network.service unit file when |
147 |
209 (or better 210) hits Gentoo. Cases like this are the use-cases for |
148 |
networkd. |
149 |
|
150 |
> 10.strange gotchas: /tmp being tmpfs using up to 50% ram. unless mounted |
151 |
> in fstab |
152 |
|
153 |
That doesn't have nothing to do with systemd: from man:mount(8): |
154 |
|
155 |
""" |
156 |
Mount options for tmpfs |
157 |
size=nbytes |
158 |
Override default maximum size of the filesystem. |
159 |
The size is given in bytes, and rounded up to entire pages. The |
160 |
default is half of the memory. The size parameter also |
161 |
accepts a suffix % to limit this tmpfs instance to that |
162 |
percentage of your physical RAM: the default, when neither size nor |
163 |
nr_blocks is specified, is size=50% |
164 |
""" |
165 |
|
166 |
systemd just mounts the tmpfs; the default *is* 50%. |
167 |
|
168 |
> 11.strange gotchas: logging is volatile by default _unless_ |
169 |
> /var/log/journal exists, when it becomes persistent due to the "auto" |
170 |
> default. |
171 |
|
172 |
It's by design; a distro can decide if it creates /var/log/journal or |
173 |
nor for its users at install time. RHEL probably will; Ubuntu |
174 |
(desktop) probably wont. |
175 |
|
176 |
> 12.transitions into systemd are non-trivial. |
177 |
|
178 |
No, they are not. |
179 |
|
180 |
> my own conclusions |
181 |
> systemd seems to be excellent for a desktop |
182 |
> good for _new_ instances of service VMs. I say new because of the large |
183 |
> job of transitioning away from openRC, but all the watchdog and better |
184 |
> resource management will help to pack datacentres. |
185 |
|
186 |
I agree. |
187 |
|
188 |
> It would also be good for big iron running many services because of this, |
189 |
> but then i thought everyone was using small fast service specific gentoo |
190 |
> VMs to compartmentalise anyway --- or was that just me? |
191 |
|
192 |
I think Gentoo users come in all sizes and shapes ;) |
193 |
|
194 |
> Unless I have completely got netctl wrong it is terrible for a |
195 |
> firewall/router scenario, or being the host server for LXC containers |
196 |
> which is a shame because resource management built in to service control |
197 |
> combined with say docker.io would be a great combination; as long as you |
198 |
> don't use custom VLAN settings. |
199 |
|
200 |
I think you got it wrong; it's just to fill the needs of simple |
201 |
networks when you don't want all the bells and whistles of a big |
202 |
network manager (not necessarily NetworkManager). |
203 |
|
204 |
BTW, Kay and Lennart Ok'd the inclusion of networkd (and they had |
205 |
thought it was missing from systemd), but it was written by Tom |
206 |
Gundersen, primarily. |
207 |
|
208 |
> As Gentoo is a meta-distro (says Larry the Cow |
209 |
> http://www.gentoo.org/main/en/about.xml) and a rolling release distro, I'm |
210 |
> all for choice, but I would sincerely hope that unlike all of the other |
211 |
> distributions from Arch to Ubuntu systemd is not adopted by default as |
212 |
> udev and baselayout transitions were bad enough. |
213 |
|
214 |
As a systemd supporter, at some point in the long, long future, I |
215 |
would be more than happy if systemd was added to the handbook as |
216 |
"secondary supported init system" in its own section. I'm completely |
217 |
fine with OpenRC as the default. |
218 |
|
219 |
Also (and I plan to work on this in the future), I would like to have |
220 |
LiveCDs and stages with systemd installed (not necessarily hosted in |
221 |
the Gentoo infrastructure), because is works really nicely install |
222 |
Gentoo from systemd-nspawn instead of a chroot. |
223 |
|
224 |
Someone would have to do that, though; I hope to help with that in the future. |
225 |
|
226 |
> I will however be installing a systemd desktop in a vm to play properly. |
227 |
> YMMV |
228 |
|
229 |
Thanks again for a succinct, impartial and objective analysis of systemd. |
230 |
|
231 |
Regards. |
232 |
|
233 |
[1] http://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/net-misc/openssh/files/sshd.service?view=markup |
234 |
[2] http://sources.gentoo.org/cgi-bin/viewvc.cgi/gentoo-x86/net-misc/openssh/files/sshd.socket?view=markup |
235 |
[3] http://cgit.freedesktop.org/systemd/systemd/tree/src/activate |
236 |
-- |
237 |
Canek Peláez Valdés |
238 |
Posgrado en Ciencia e Ingeniería de la Computación |
239 |
Universidad Nacional Autónoma de México |