array of IPv4 addresses. BTW the Kea definition does
not catch the fact addresses come by destination+gateway pairs but
this should not be a problem...
Regards
Francis Dupont <fdup...@isc.org>
PS: this is no easy when to express this kind of constraints and BTW
it i
timer values are configured
and a default value copied from the global scope is not configured?
Regards
Francis Dupont <fdup...@isc.org>
PS: if the action of deriveParams (src/lib/cc/simple_parser.h) is misleading
we have somewhere the code of a factorParams doing the opposite.
It was
I gave the ticket #4539 "kea-dhcp6 must join All_DHCP_Servers too"
another chance...
Regards
Francis Dupont <fdup...@isc.org>
PS: I put it on the kea-proposed milestone so it will be put in
Kea 1.3 final (i.e., addressed before 1.3 release), Kea 1.4 or
back in Out
"test": "relay4[2].text == '100.101.101.0'"
=> please try "test": "relay4[2].hex == 100.101.101.0"
(.text does not return what you want if the (sub-)option is not a string
and IP address litterals are supported).
A client class is the easiest way to perform specific subnet selection.
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
IMHO the simplest is to write a little hook which resets the giaddr,
or to understand why the client or relay sets it and fix this.
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.i
Does it not enough to define it globally?
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
Jason Lixfeld writes:
> There a translation tool to migrate ISC DHCP configs to Kea?
=> we are working on one but I don't know how it will be made available
(e.g. assisted web-based service?).
Regards
Francis Dupont <fdup...@isc.org>
_
config the tool can't infer the
interface list...
Regards
Francis Dupont <fdup...@isc.org>
PS: in Jeff's case it is better to select subnets using relay addresses
than using the receiving interface because a priori relays serve
different physical subnets so should no
A client-class in a subnet may be used only to refuse a subnet selection.
In your case your config requires a positive selector, for instance
a relay address (relayed messages) or an interface (direct connected
clients).
Regards
Francis Dupont <fdup...@isc.org>
PS: there is a third possi
itay cohen writes:
> its kinda working,
> after changing the address i'm getting this error and kea stops sending
> responses.
>
> DHCP4_PACKET_SEND_FAIL [hwtype=1 00:0c:29:45:76:07],
> cid=[01:00:0c:29:45:76:07], tid=0xdbed5a23: failed to send DHCPv4 packet:
> pkt4 send failed: sendmsg()
> does some now know how to change the source ip of dhcp messages with a hook
> ?
=> setRemoteAddr(newaddr) on the received packet should do this...
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@list
't listen on 2001:db8:beef:55::2 by default.
> The issue is very strange that is why I need your help. The strange thing
> is that perfdhcp worked perfectly for ipv4.
=> DHCPv4 uses (by default, you can change the service socket type)
a BPF/LPF/etc socket, DHCPv6 uses standard UDP.
configuration in DHCPv4 server guide).
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
h information. But maybe there is
> some trick I can use right now?
=> in both cases I am afraid you need a hook. BTW there was a hook calling
scripts announced in this list some days ago so perhaps you do not have
to write a line of C++ (*) code.
Regards
Francis Dupont <fd
Jason Lixfeld writes:
> However, running -t said syntax was OK.
=> -t does not check everything, for instance hook parameters.
Regards
Francis Dupont <fdup...@isc.org>
BTW the manual page says that "hook libraries are not loaded" so
don't expect to have a not loaded
Francis Dupont writes:
I should have added my answer applies only to static (i.e. in config file)
definitions / setting data values.
> Marcin Jurczuk writes:
> > I'm evaluating isc-dhcp replacement with kea and I've stumbled upon some co
> nf
> > iguration options tha
he option is standard
for ISC DHCP (defined in common/tables.c) we have 2 tickets which
implement it in Kea (if it is not yet supported).
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.
) you have a work-around by using
a binary value: "csv-format": false, "data": "0123456789abcdef".
Regards
Francis Dupont <fdup...@isc.org>
PS: I can translate your ISC DHCP config into Kea but it won't change
the fact that multiple enterprises are not suppo
Try to get the HOOKS_OPEN_ERROR which returns errors from dlopen().
Note the main reason to get dlopen to not find DSOs is to forget to
set LD_LIBRARY_PATH...
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.i
"Chaigneau, Nicolas" writes:
> I assume the file is looked for in the same directory as the including file?
> ?
>
> Can you also specify absolute file paths ? ( ?>)
> And relative paths ? ()
=> it does fopen(filename.c_str(), "r") so the answer is yes t
;option[55].hex == 0x0103072B".
> This loads correctly on service start but I don't think I've gotten it to
> successfully match to a client yet.
=> you can debug the classification evaluation.
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
=> yes, valid-lifetime, renew-timer and rebind-timer are supported
parameters in DHCPv4 subnets.
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
easy to fix.
BTW if you can send to us all details to help in the solution definition?
Thanks
Francis Dupont <fdup...@isc.org>
PS: the only round-around is to define in src/lib/dhcp/std_option_defs.h
the option 43 to binary without encapsulated space
User contexts are strictly more powerful as they are part of objects so
can be retrieve by hooks. They just need to be more general, i.e.
current user-context support is still a (successful) experimentation.
Regards
Francis Dupont <fdup...@isc.
James Sumners writes:
> Unfortunately the parser doesn't ignore unknown properties.
=> not unfortunately: it is by design and I am sure you'd like the
parser to catch a trivial spelling error than to silently ignore it.
Regards
Francis Dupont <fdup...@isc.org>
PS: as you expect
n is fine but as you found it can handle only one route
with a subnet which encodes into 3 bytes...
Unfortunately there is only one correct definition: say the layout
is binary (BTW it is the default for unknown options) and encode the
value manually...
Regards
Francis Dupont <fdup...@isc.or
long hook startup you should initiate it
when the hook is loaded (unfortunately today a hook is loaded more than
once?)
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
l a lot of dhcp-packets :p)
>
> Any tip, trick, configuration or whatever on that issue ?
=> slow processing is incompatible with the DHCP protocol so basically
there is no final fix, only tricks to make things less broken.
Regards
Francis Dupont <fdup...@isc.org>
___
somewhere (you are not the first asking this).
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
idea).
Thanks
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
?
=> which system? The code sets the field to IPTOS_LOWDELAY for BPF & co,
and does not touch it for UDP (DHCPv6 default, DHCPv4 when dhcp socket
type is configured). This does not seem crazy and for sure should not
raise any problem...
Regards
Francis Dupont <fdup...@isc.org>
PS: IS
I don't think Kea does something special with fe80:: but kernels should
as it is the any router anycast address, so illegal as a source address...
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
er detail
which it is (too) easy to get wrong. I recommend to use wireshark or
tcpdump to inspect packets on the wire, in particular the PRL in
the discover and options in the offer.
Regards
Francis Dupont <fdup...@isc.org>
PS: option 150 got multiple definitions in the past so is *not* a
hat some critical values are not 0, for instance
the result of the dynamic cast. This should make your code more robust.
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
to write a hook which
patches the PRL (option 55 dhcp-parameter-request-list) as this feature
is implemented in ISC DHCP (it will be more user friendly in future Kea).
Thanks
Francis Dupont <fdup...@isc.org>
PS: from RFC 2132:
9.8. Parameter Request List
This option is u
If you use the option the easiest is to define a global option-data.
BTW there is a ticket to allow bootp parameters in more contexts.
Regards
Francis Dupont <fdup...@isc.org>
PS: option-data can be global, per subnet, per reservation, per class
and since yesterday per pool. Note client
ything after
is ignored... So there is *nothing* else in RFCs about option order
and as far as I know it is at implementor choice (which is different
for ISC DHCP and Kea BTW).
Thanks
Francis Dupont <fdup...@isc.org>
PS: you can use a hook to pack options in your
I meant that each time I had a choice between clang and gcc these last
years clang was better.
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
the Makefile or to install
the library manually (first option is more complex but it should make
things right).
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
pkt->addClass("my-client-class"); should work well. BTW the client class
include in dhcp/classify.h
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
V8 API is unstable so it is hard to maintain a code written in v8.
At least I am sure the v8 API version I used last year is today enough
different than some part of the code need to be rewritten.
Regards
Francis Dupont <fdup...@isc.org>
PS: and a programming language is not popular b
I did an experiment writing hooks in other languages than C++ and
Python was in the list. I committed this branch fdxhook to github,
things (including the doc you should read) are under src/hooks/external
(based in an old version of Kea as this is nearly one year old).
Regards
Francis Dupont
- if libssl.so is not listed by ldd then add a hack at the end of
src/lib/cryptolink/openssl_link.cc with a reference to a libssl.so
symbol. Choose one (e.g., SSL_library_init), add its include
() and a call to it as in lexer.ll. Note the function
will be never really called but it will be r
I suggest to add
"interface": "ens192",
into the "subnet": "172.0.1.0/24" specification if the failing client
is directly attached to this subnetwork.
Regards
Francis Dupont <fdup...@isc.org>
Yury Tsaregorodtsev writes:
>
>
> --===
close a partial list or map using
an include) so do not touch the syntax.
Thanks
Francis Dupont <fdup...@isc.org>
PS: the main inconvenient from the strict no trailing comma rule is
when you put some part into a comment, in particular when generating
the config...
_
r as I know the only required parameter in a subnet is the
"subnet" range. Pools are not required and without a pool there is no
available address so only known hosts which have a reservation can
get an address. This is a theory, i.e. you have to try it, but on
the paper it does exactly what you
Stas Starikevich writes:
> Can someone suggest a way to serve only white-listed (with host reservation)
> clients?
> I.e. if client is not in the 'hosts' table then it shouldn't not get any leas
> es (and boot from the HDD, but not from PXE).
=> did you try with no pool?
Regards
ks
are doing the same thing: they print the query using Kea C++ toText method
(kea-pkt4.str in python). They are just to show it is feasible and
to find all problems, not to do real things (even they are a good
start for a hook library with the same power than C++).
Regards
Francis Dupont <
Joelson Vendramin writes:
> Please let me know when you have this "flex-id" implemented!
=> it was a hour ago.
Thanks
Francis Dupont <fdup...@isc.org>
PS: it is a premium feature (Tomek shall explain what this is meant).
___
Known and fixed defect:
1213. [bug] fdupont
Option string values containing comma can now be specified
correctly by preceding comma with double backslashes (e.g.
"foo\\,bar").
(Trac #5105, git fa79ac2396aa94d7bac91bd12d3593ebaaa9386d)
Regard
ot
compressed (compression is explicitly allowed in this option
which BTW has for format an array of DNS names).
The DNS names (aka FQDNs) in wired format are supposed to be supported
by Kea but it seems only DHCPv6 standard options use it. So please
open a
Look at the mac-sources stuff in the Kea user guide.
Regards
Francis Dupont <fdup...@isc.org>
Rui Pedro Caldeira writes:
> --===4576728167823046396==
> Content-Language: en-US
> Content-Type: multipart/alternative;
&
I chose the escape because you can get embedded commas inside a field,
i.e., the string value is just a particular case (even it seems the most
current).
Thanks
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
Kea is still young so it has no to support a long history as bind or
ISC DHCP. Or with other words all Kea releases are major releases...
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.i
uot;data\": \"foo,bar\" in
the case you read the patch content).
BTW something I expect from the reviewer is a request for documentation
so if you have an idea about to explain the escape feature in the Kea
user guide...
Regards
Francis Dupont <fdup...@isc.org>
PS: to
Currently in Kea you can't put a comma (character ',') in an option string
value. It is a known problem (ticket #5105) but its solution was not
yet reviewed... Note you can still use a binary value which has to
this problem.
Regards
Francis Dupont <fdup...@isc.org>
PS: If you'd like
are not late so IMHO this feature should be
supported pretty quickly.
Thanks
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
IMHO what you need is a tool to translate ISC DHCP lease database into
a Kea one. This seems a good idea for the day we'll work on migration
tools.
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
>>* Francis Dupont <fdup...@isc.org>:
>> In ISC DHCP the check of the server ID (DHCPv4 option 54) is optional
>> (server-id-check dhcpd.conf option, default no check).
> Can it be made optional in KEA as well?
=> ISC DHCP manual has an extensive discu
In ISC DHCP the check of the server ID (DHCPv4 option 54) is optional
(server-id-check dhcpd.conf option, default no check).
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/m
SC DHCP have command line options to not daemonize. Kea servers
have none but don't daemonize (i.e., they stay in foreground without
closing standard in, out and error, nor create a new login session.
So I don't understand what is your problem? (i.e., IMHO there is
nothing to solve).
Regards
Francis
We know the address given is server-identifier should get more control,
there are some tickets about this (and as far as I know no time to
deal with them soon...).
Regards
Francis Dupont <fdup...@isc.org>
Igor Smitran writes:
> Hi list,
>
> Did any of you tried to set custom se
delayed ACKs). Another point is Kea is currently not multi-threaded
(again DHCPv4 over DHCPv6 is an exception because it uses 2 processes,
one managing the DHCPv6 part, the other the DHCPv4 part).
Regards
Francis Dupont <fdup...@isc.org>
PS: about your example you have a trade-off between rel
If nobody answers before Monday I'll look at Jenkins configs but
I believe the corresponding make commands are explicitly called.
I apologize but my phone is ringing...
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea
=> I believe you checked with tcpdump or wireshark the solicit was
received by the server system (BTW a capture could be useful).
Did your disabled the IPv6 firewall? (the question is a bit silly
but this point is easy to forget, myself I was caught from time to
time).
Regards
Francis Dup
server-id 0.0.0.0 is considered as invalid by only some clients
but not all... Anyway there are at least 2 tickets (#3280 and
#4063) asking for a configurable server-id...
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea
someone already asked please
ignore: I am back from holidays)
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
mp; paste the code after this comment... Or if
the other specific processing is useful too puts all clients in the
right class.
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
Downloaded the ports (vs packages, cf my FreeBAS comments) for OpenBSD:
it has a net/kea directory.
Regards
Francis Dupont <fdup...@isc.org>
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
there is no net/kea in it (yet).
Regards
Francis Dupont <fdup...@isc.org>
PS: I know some people in the cited *BSD teams if needed.
___
Kea-users mailing list
Kea-users@lists.isc.org
https://lists.isc.org/mailman/listinfo/kea-users
Did you try to add an address to the interface in the kea.conf?
>From time to time it makes things magically working (at least for me).
Regards
Francis Dupont <fdup...@isc.org>
PS: http://git.kea.isc.org/~tester/kea/guide/kea-guide.html#dhcp6-unicast
PPS: your tcpdump shows a link loc
201 - 272 of 272 matches
Mail list logo