Sender | Message | Time |
---|---|---|
24 Nov 2023 | ||
nakof_75462 joined the room. | 14:24:47 | |
25 Nov 2023 | ||
arta.eth joined the room. | 03:25:25 | |
quantumcat40 joined the room. | 13:47:17 | |
27 Nov 2023 | ||
gingimli | 21:01:29 | |
29 Nov 2023 | ||
gamba47 joined the room. | 13:59:58 | |
30 Nov 2023 | ||
wowbagger_40528 joined the room. | 10:49:13 | |
hofione | hi, i cannot decide what is the normal behavior, so asking instead i have 2 syslog() sourcesone UDP and one TCP
(please note the no-parse flags)the UDP one eats the none RFC formatted input the TCP one reports invalid frame header test input was fed by
where the input file content was
which one is correct? UDP source or TCP source behavior? (my bet, as both the source and the loggen got no-parse options, the UDP one is the correct) | 11:56:33 |
hofione | * hi, i cannot decide what is the normal behavior, so asking instead i have 2 syslog() sourcesone UDP and one TCP
(please note the no-parse flags)the UDP one eats the none RFC formatted input the TCP one reports invalid frame header test input was fed by
where the input file content was
which one is correct? UDP source or TCP source behavior? (my bet, as both the source and the loggen got no-parse options, the UDP one is the correct) | 12:01:38 |
gamba47 | 12:52:03 | |
gabor_88547 joined the room. | 18:14:19 | |
calmazans joined the room. | 20:13:08 | |
1 Dec 2023 | ||
haxodon_82096 joined the room. | 09:12:35 | |
4 Dec 2023 | ||
pilosula_32045 joined the room. | 07:10:12 | |
6 Dec 2023 | ||
Mark joined the room. | 01:09:09 | |
hofione | bazsi77 overorion could you please take a look at this? https://github.com/syslog-ng/syslog-ng/pull/4544 the actual solution gives a generic tcp proxy that can be used for all the tcp based sources e.g. network(), syslog() i've tested it with the network() source that seems to be fine for me but i couldn't find any RFC of transmitting syslog over a proxy RFC 6587 is only about the TCP part so i do not know if it's a valid case at all now if it was valid then i would need some help as there are/might be multiple issues (e.g. in my interpretation the proxy header should not be octet counted that the framed proto is not prepared for in its actual state) (actually, my tests are passed on syslog() as well for all the proxied variants, but only with well prepared inputs a.k.a octet counted proxy headers) | 11:59:15 |
hofione | * bazsi77 overorion could you please take a look at this? https://github.com/syslog-ng/syslog-ng/pull/4544 the actual solution gives a generic tcp proxy that can be used for all the tcp based sources e.g. network(), syslog() i've tested it with the network() source that seems to be fine for me but i couldn't find any RFC of transmitting syslog over a proxy RFC 6587 is only about the TCP part so i do not know if it's a valid case at all now if it was valid then i would need some help as there are/might be multiple issues (actually, my tests are passed on syslog() as well for all the proxied variants, but only with well prepared inputs a.k.a octet counted proxy headers) | 12:35:11 |
hofione | * bazsi77 overorion could you please take a look at this? https://github.com/syslog-ng/syslog-ng/pull/4544 the actual solution gives a generic tcp proxy that can be used for all the tcp based sources e.g. network(), syslog() i've tested it with the network() source that seems to be fine for me but i couldn't find any RFC of transmitting syslog over a proxy RFC 6587 is only about the TCP part so i do not know if it's a valid case at all now (actually, my tests are passed on syslog() as well for all the proxied variants, but only with well prepared inputs) | 12:36:11 |
steveizzle joined the room. | 16:56:19 | |
bazsi77 | Of course I'd look at it. Bit overwhelmed these days. | 19:50:01 |
bazsi77 | I am not sure I understand why LogTransportSocketProxy cannot be a LogTransport instance. That's the design that I had in mind originally. Maybe multitransport does make that impossible? I'll check this in more details | 19:56:21 |
7 Dec 2023 | ||
overorion changed their display name from UBWGUNNER to overorion. | 07:09:58 | |
overorion | Not sure about this week (really occupied these days), but next week I will have a look, thanks! | 07:09:58 |
mstopa joined the room. | 11:04:48 | |
hofione | as i can see, it cannot be LogTransport based (simply) because in that case, either 1.) it must extend (or become part of the) LogTransport, and all of the inherited classes (MultiTransport, LogTransportSocket, etc.) must be inherited from the new proxied transport class, or 2.) we must add all the LogTransport proxied variants as well (e.g. ProxiedMultiTransport, ProxiedLogTransportSocket, etc.) that we surely do not want the solution i've added seemed to be better for me, than option 1.) above, yes it also has PROs and CONs, this is why I asked support earlier, to discuss what would be the best solution here | 11:12:30 |
hofione | * as i can see, it cannot be LogTransport based (simply) because in that case, either 1.) it must extend (or become part of the) LogTransport, and all of the inherited classes (MultiTransport, LogTransportSocket, etc.) must be inherited from the new proxied transport class, or 2.) we must add all the LogTransport proxied variants as well (e.g. ProxiedMultiTransport, ProxiedLogTransportSocket, etc.) that we surely do not want the solution i've added seemed to be better for me than option 1.) above, yes it also has PROs and CONs, this is why I asked support earlier, to discuss what would be the best solution here | 11:15:19 |
bazsi77 | I am working on fixing clang support and get back to this. | 11:15:33 |
hofione | yepp CzP mentioned, I'm curious how it will end, I started once for a short time, with no luck (bunch of c++ issues) | 11:16:44 |
bazsi77 | It already works more or less. | 11:16:59 |
hofione | * yepp CzP mentioned it, I'm curious how it will end, I started once for a short time, with no luck (bunch of c++ issues) | 11:17:03 |
bazsi77 | I found a glib bug but it compiles for me on Linux with clang/clang++ | 11:17:41 |