this is bugging me for at least one year now. every time I have to merge a new packet it takes for ever to get the archive from the main distfiles website. this is due to the fact that your ipv6 roudrobin always seems to have a dead host in it and my RNG always seems to favor that particular mirror. so please be so kind and set up a cron-based script that tests the connectivity and removes the dead entries from that record. lipo ~ # dig +short AAAA distfiles.gentoo.org 2605:bc80:3010::134 2600:3402:200:227::2 2600:3404:200:237::2 2a00:8a60:e012:a00::21 lipo ~ # telnet 2605:bc80:3010::134 80 Trying 2605:bc80:3010::134... Connected to 2605:bc80:3010::134. Escape character is '^]'. ^] telnet> quit Connection closed. lipo ~ # telnet 2600:3404:200:237::2 80 Trying 2600:3404:200:237::2... Connected to 2600:3404:200:237::2. Escape character is '^]'. ^] telnet> quit Connection closed. lipo ~ # telnet 2a00:8a60:e012:a00::21 80 Trying 2a00:8a60:e012:a00::21... telnet: connect to address 2a00:8a60:e012:a00::21: Connection timed out lipo ~ # telnet 2600:3402:200:227::2 80 Trying 2600:3402:200:227::2... Connected to 2600:3402:200:227::2. Escape character is '^]'. ^] telnet> quit Connection closed.
(In reply to petre rodan from comment #0) > this is bugging me for at least one year now. every time I have to merge a > new packet it takes for ever to get the archive from the main distfiles > website. > this is due to the fact that your ipv6 roudrobin always seems to have a dead > host in it and my RNG always seems to favor that particular mirror. I'm happy to remove any unavailable mirrors. > > so please be so kind and set up a cron-based script that tests the > connectivity and removes the dead entries from that record. Portage should be retrying a number of times; the AAAA record has multiple entries precisely so that if one is dead portage will try another one. While this might add latency for downloads (each attempt has some time cost) we shouldn't be in a state where more than say, 3 of the records are dead; and so retries should cover us here. Cron based DNS edits are non-trivial in our existing setup and have their own issues that I think outweigh the gains of automation here. > > lipo ~ # dig +short AAAA distfiles.gentoo.org > 2605:bc80:3010::134 > 2600:3402:200:227::2 > 2600:3404:200:237::2 > 2a00:8a60:e012:a00::21 > > lipo ~ # telnet 2605:bc80:3010::134 80 > Trying 2605:bc80:3010::134... > Connected to 2605:bc80:3010::134. > Escape character is '^]'. > ^] > telnet> quit > Connection closed. > > lipo ~ # telnet 2600:3404:200:237::2 80 > Trying 2600:3404:200:237::2... > Connected to 2600:3404:200:237::2. > Escape character is '^]'. > ^] > telnet> quit > Connection closed. > > lipo ~ # telnet 2a00:8a60:e012:a00::21 80 > Trying 2a00:8a60:e012:a00::21... > telnet: connect to address 2a00:8a60:e012:a00::21: Connection timed out > lipo ~ # telnet 2600:3402:200:227::2 80 > Trying 2600:3402:200:227::2... > Connected to 2600:3402:200:227::2. > Escape character is '^]'. > ^] > telnet> quit > Connection closed.
(In reply to Alec Warner from comment #1) > (In reply to petre rodan from comment #0) > > this is bugging me for at least one year now. every time I have to merge a > > new packet it takes for ever to get the archive from the main distfiles > > website. > > this is due to the fact that your ipv6 roudrobin always seems to have a dead > > host in it and my RNG always seems to favor that particular mirror. > > I'm happy to remove any unavailable mirrors. > > > > > so please be so kind and set up a cron-based script that tests the > > connectivity and removes the dead entries from that record. > > Portage should be retrying a number of times; the AAAA record has multiple > entries precisely so that if one is dead portage will try another one. While > this might add latency for downloads (each attempt has some time cost) we > shouldn't be in a state where more than say, 3 of the records are dead; and > so retries should cover us here. > > Cron based DNS edits are non-trivial in our existing setup and have their > own issues that I think outweigh the gains of automation here. > > > > > lipo ~ # dig +short AAAA distfiles.gentoo.org > > 2605:bc80:3010::134 > > 2600:3402:200:227::2 > > 2600:3404:200:237::2 > > 2a00:8a60:e012:a00::21 > > > > lipo ~ # telnet 2605:bc80:3010::134 80 > > Trying 2605:bc80:3010::134... > > Connected to 2605:bc80:3010::134. > > Escape character is '^]'. > > ^] > > telnet> quit > > Connection closed. > > > > lipo ~ # telnet 2600:3404:200:237::2 80 > > Trying 2600:3404:200:237::2... > > Connected to 2600:3404:200:237::2. > > Escape character is '^]'. > > ^] > > telnet> quit > > Connection closed. > > > > lipo ~ # telnet 2a00:8a60:e012:a00::21 80 > > Trying 2a00:8a60:e012:a00::21... > > telnet: connect to address 2a00:8a60:e012:a00::21: Connection timed out > > lipo ~ # telnet 2600:3402:200:227::2 80 > > Trying 2600:3402:200:227::2... > > Connected to 2600:3402:200:227::2. > > Escape character is '^]'. > > ^] > > telnet> quit > > Connection closed. Note that for me, all 4 of these are working currently. -A
2a00:8a60:e012:a00::21 is not working for me and it looks like maybe I get filtered at the very last hop. my local ipv6 is currently 2a02:2f0e:808:4e10:1687:67f5:603f:61c1/64 . prodan@lipo ~ $ telnet 2a00:8a60:e012:a00::21 80 Trying 2a00:8a60:e012:a00::21... telnet: connect to address 2a00:8a60:e012:a00::21: Connection timed out prodan@lipo ~ $ traceroute6 2a00:8a60:e012:a00::21 traceroute to 2a00:8a60:e012:a00::21 (2a00:8a60:e012:a00::21), 30 hops max, 80 byte packets 1 2a02:2f0e:808:4e10::1 (2a02:2f0e:808:4e10::1) 4.076 ms 4.012 ms 3.973 ms 2 2a02:2f0e:8ff:ff00::2 (2a02:2f0e:8ff:ff00::2) 16.673 ms 15.185 ms 15.133 ms 3 2a02:2f0e:eff:ff01::1 (2a02:2f0e:eff:ff01::1) 17.230 ms 16.450 ms 16.385 ms 4 2a02:2f00:8708:2:2:0:6:0 (2a02:2f00:8708:2:2:0:6:0) 11.016 ms 11.879 ms 12.977 ms 5 rcsrds-ic-323846-ffm-b4.c.telia.net (2001:2000:3080:131e::2) 35.586 ms 40.588 ms 35.500 ms 6 * * * 7 kr-aah15.x-win.dfn.de (2001:638:c:a100::2) 33.511 ms 34.700 ms 34.794 ms 8 fw-xwin-2-vlan106.noc.rwth-aachen.de (2a00:8a60:0:f000::12) 32.409 ms 32.387 ms 33.379 ms 9 n7k-lssnord-1-vl158.noc.rwth-aachen.de (2a00:8a60:0:f001::4) 35.245 ms 37.077 ms 35.676 ms 10 n7k-sw23-2-et2-1.noc.rwth-aachen.de (2a00:8a60:0:f025::2) 38.246 ms 36.529 ms 38.081 ms 11 * * * 12 * * * 13 * * * 14 * * * 15 * * * 16 * * * 17 * * * 18 * * * 19 * * * 20 * * * 21 * * * [..] from a different location that uses a Hurricane Electric tunnel the mirror is indeed reached: skycaves ~ # telnet 2a00:8a60:e012:a00::21 80 Trying 2a00:8a60:e012:a00::21... Connected to 2a00:8a60:e012:a00::21. Escape character is '^]'. ^] telnet> quit Connection closed. skycaves ~ # traceroute6 2a00:8a60:e012:a00::21 traceroute to 2a00:8a60:e012:a00::21 (2a00:8a60:e012:a00::21), 30 hops max, 80 byte packets 1 petrerodan-1.tunnel.tserv5.lon1.ipv6.he.net (2001:470:1f08:34a::1) 5.089 ms 8.191 ms 11.104 ms 2 10ge3-16.core1.lon2.he.net (2001:470:0:67::1) 11.021 ms 10.993 ms 10.970 ms 3 100ge7-1.core1.fra1.he.net (2001:470:0:37::2) 23.022 ms 22.999 ms 22.977 ms 4 * * * 5 kr-aah15.x-win.dfn.de (2001:638:c:a100::2) 27.421 ms 27.082 ms 27.927 ms 6 fw-xwin-1-vlan106.noc.rwth-aachen.de (2a00:8a60:0:f000::11) 27.031 ms 17.638 ms 17.575 ms 7 n7k-lssnord-1-vl158.noc.rwth-aachen.de (2a00:8a60:0:f001::4) 18.879 ms 18.342 ms 18.162 ms 8 n7k-sw23-2-et2-1.noc.rwth-aachen.de (2a00:8a60:0:f025::2) 18.485 ms 19.092 ms 18.979 ms 9 ftp.halifax.rwth-aachen.de (2a00:8a60:e012:a00::21) 17.461 ms 17.776 ms 17.438 ms it is true that a timeout exists, but for a packet like dev-texlive/texlive-fontsextra where hundreds of files need to be downloaded it will take 60 seconds of timeout * 3 retries * hundreds of files = many hours of nothing. and I swear that the broken ip is always the one that ends up used first - which I find extremely weird. >>> Downloading 'http://distfiles.gentoo.org/distfiles/texlive-module-Asana-Math-2017.tar.xz' --2019-03-30 07:13:11-- http://distfiles.gentoo.org/distfiles/texlive-module-Asana-Math-2017.tar.xz Resolving distfiles.gentoo.org... 2a00:8a60:e012:a00::21, 2600:3402:200:227::2, 2605:bc80:3010::134, ... Connecting to distfiles.gentoo.org|2a00:8a60:e012:a00::21|:80... zzZZZzzzZZ ^C I did indeed tweak the local FETCHCOMMAND, but that's not a proper fix.
(In reply to petre rodan from comment #3) > 2a00:8a60:e012:a00::21 is not working for me and it looks like maybe I get > filtered at the very last hop. my local ipv6 is currently > 2a02:2f0e:808:4e10:1687:67f5:603f:61c1/64 . Ok, well lets follow up with them to see why you appear to be filtered. My original test was HE, but I also found some other ipv6 native hosts and I tried those and they worked; so we should determine why your reachability is impaired to this particular mirror. I'm not sure we should remove it just because 1 user reports a reachability problem when its likely that many users are using the mirror just fine. > > prodan@lipo ~ $ telnet 2a00:8a60:e012:a00::21 80 > Trying 2a00:8a60:e012:a00::21... > telnet: connect to address 2a00:8a60:e012:a00::21: Connection timed out > > prodan@lipo ~ $ traceroute6 2a00:8a60:e012:a00::21 > traceroute to 2a00:8a60:e012:a00::21 (2a00:8a60:e012:a00::21), 30 hops max, > 80 byte packets > 1 2a02:2f0e:808:4e10::1 (2a02:2f0e:808:4e10::1) 4.076 ms 4.012 ms 3.973 > ms > 2 2a02:2f0e:8ff:ff00::2 (2a02:2f0e:8ff:ff00::2) 16.673 ms 15.185 ms > 15.133 ms > 3 2a02:2f0e:eff:ff01::1 (2a02:2f0e:eff:ff01::1) 17.230 ms 16.450 ms > 16.385 ms > 4 2a02:2f00:8708:2:2:0:6:0 (2a02:2f00:8708:2:2:0:6:0) 11.016 ms 11.879 > ms 12.977 ms > 5 rcsrds-ic-323846-ffm-b4.c.telia.net (2001:2000:3080:131e::2) 35.586 ms > 40.588 ms 35.500 ms > 6 * * * > 7 kr-aah15.x-win.dfn.de (2001:638:c:a100::2) 33.511 ms 34.700 ms 34.794 > ms > 8 fw-xwin-2-vlan106.noc.rwth-aachen.de (2a00:8a60:0:f000::12) 32.409 ms > 32.387 ms 33.379 ms > 9 n7k-lssnord-1-vl158.noc.rwth-aachen.de (2a00:8a60:0:f001::4) 35.245 ms > 37.077 ms 35.676 ms > 10 n7k-sw23-2-et2-1.noc.rwth-aachen.de (2a00:8a60:0:f025::2) 38.246 ms > 36.529 ms 38.081 ms > 11 * * * > 12 * * * > 13 * * * > 14 * * * > 15 * * * > 16 * * * > 17 * * * > 18 * * * > 19 * * * > 20 * * * > 21 * * * > [..] > > from a different location that uses a Hurricane Electric tunnel the mirror > is indeed reached: > > skycaves ~ # telnet 2a00:8a60:e012:a00::21 80 > Trying 2a00:8a60:e012:a00::21... > Connected to 2a00:8a60:e012:a00::21. > Escape character is '^]'. > ^] > telnet> quit > Connection closed. > > skycaves ~ # traceroute6 2a00:8a60:e012:a00::21 > traceroute to 2a00:8a60:e012:a00::21 (2a00:8a60:e012:a00::21), 30 hops max, > 80 byte packets > 1 petrerodan-1.tunnel.tserv5.lon1.ipv6.he.net (2001:470:1f08:34a::1) > 5.089 ms 8.191 ms 11.104 ms > 2 10ge3-16.core1.lon2.he.net (2001:470:0:67::1) 11.021 ms 10.993 ms > 10.970 ms > 3 100ge7-1.core1.fra1.he.net (2001:470:0:37::2) 23.022 ms 22.999 ms > 22.977 ms > 4 * * * > 5 kr-aah15.x-win.dfn.de (2001:638:c:a100::2) 27.421 ms 27.082 ms 27.927 > ms > 6 fw-xwin-1-vlan106.noc.rwth-aachen.de (2a00:8a60:0:f000::11) 27.031 ms > 17.638 ms 17.575 ms > 7 n7k-lssnord-1-vl158.noc.rwth-aachen.de (2a00:8a60:0:f001::4) 18.879 ms > 18.342 ms 18.162 ms > 8 n7k-sw23-2-et2-1.noc.rwth-aachen.de (2a00:8a60:0:f025::2) 18.485 ms > 19.092 ms 18.979 ms > 9 ftp.halifax.rwth-aachen.de (2a00:8a60:e012:a00::21) 17.461 ms 17.776 > ms 17.438 ms > > it is true that a timeout exists, but for a packet like > dev-texlive/texlive-fontsextra where hundreds of files need to be downloaded > it will take 60 seconds of timeout * 3 retries * hundreds of files = many > hours of nothing. and I swear that the broken ip is always the one that ends > up used first - which I find extremely weird. > > >>> Downloading 'http://distfiles.gentoo.org/distfiles/texlive-module-Asana-Math-2017.tar.xz' > --2019-03-30 07:13:11-- > http://distfiles.gentoo.org/distfiles/texlive-module-Asana-Math-2017.tar.xz > Resolving distfiles.gentoo.org... 2a00:8a60:e012:a00::21, > 2600:3402:200:227::2, 2605:bc80:3010::134, ... > Connecting to distfiles.gentoo.org|2a00:8a60:e012:a00::21|:80... zzZZZzzzZZ > ^C > > I did indeed tweak the local FETCHCOMMAND, but that's not a proper fix.
We moved distfiles.gentoo.org to a CDN, my hope is you have better connectivity to the CDN.