Message-ID: <682e5d6d@news.ausics.net> From: not@telling.you.invalid (Computer Nerd Kev) Subject: Re: Anybody Using IPv6? Newsgroups: comp.os.linux.misc References: <100hijb$25ta5$5@dont-email.me> <100hl34$3of0i$1@news1.tnib.de> <100i89e$19k3$1@news1.tnib.de> <100i92h$2a8rb$3@dont-email.me> <100ieve$1o3d$1@news1.tnib.de> <100irbu$2j9c$1@news1.tnib.de> <100jqfp$4o04$1@news1.tnib.de> <100k1v0$2o6n2$1@dont-email.me> <100kcnb$649q$1@news1.tnib.de> <100kpc3$6vge$1@news1.tnib.de> User-Agent: tin/2.0.1-20111224 ("Achenvoir") (UNIX) (Linux/2.4.31 (i586)) NNTP-Posting-Host: news.ausics.net Date: 22 May 2025 09:10:37 +1000 Organization: Ausics - https://newsgroups.ausics.net Lines: 52 X-Complaints: abuse@ausics.net Path: ...!eternal-september.org!feeder3.eternal-september.org!news.bbs.nz!news.ausics.net!not-for-mail Bytes: 3486 Carlos E. R. wrote: > On 2025-05-21 16:52, Marc Haber wrote: >> "Carlos E. R." wrote: >>> On 2025-05-21 13:16, Marc Haber wrote: >>>> I think that the local resolver should? also refrain from asking for >>>> AAAA records if the local system doesn't have IPv6, but I don't know >>>> whether this special-case handling is implemented at all. And I'm too >>>> lazy to look that up. >>>> >>>> But all this needs to be taken into account before someone can comment >>>> about speed of one IP protocol compared to the other on a level that >>>> is beyond passing myths. >>> >>> It just is a perceived fact. On some machines, if an application gets >>> back from the system a list of addresses to try, and tries IPv6 first >>> when there is no actual IPv6 internet connectivity, there is a small >>> delay waiting for the request to fail, and then try the next address in >>> the list. >> >> pcap/strace or it didn't happen. > > Asking for a trace _now_ is ridiculous. It certainly did happen, and to > to several people. Years ago. Looks like it was only last year when I encountered package list downloads failing in Aptitude on an IPv4-only VPS due to it trying to connect on IPv6. Disabling IPv6 on there made perfect sense - it's intended to be a stable system, not a testing ground for applications. I knew IPv6 wasn't available, and I'd seen such behaviour before in an unimportant program on another system, so really it's my fault for leaving the door open to such bugs by pointlessly leaving the kernel's IPv6 support enabled. >> Please note that _broken_ IPv6, for example when the router announces >> an IPv6 but the network doesn't return a host unreachable ICMPv6 >> message from the place where connectivity is missing, will cause an >> IPv6-enabled application to wait for the time out. But that is an >> error in the _network_ setup, and should not happen in the case where >> the end system (the one with the application running) has v6 enabled >> on a non-v6-enabled network. But if that network error has been made, you'll avoid trouble if IPv6 is disabled. So if you know IPv6 isn't available anyway, there's a clear advantage to disabling it in the kernel and dodging these potential sources of failure even if they "should not happen". -- __ __ #_ < |\| |< _#