)]}'
{"neutron/agent/linux/dhcp.py":[{"author":{"_account_id":1131,"name":"Brian Haley","email":"haleyb.dev@gmail.com","username":"brian-haley"},"change_message_id":"9ead182c657b629fa30993dd40b5551131117152","unresolved":false,"context_lines":[{"line_number":328,"context_line":"            \u0027dnsmasq\u0027,"},{"line_number":329,"context_line":"            \u0027--no-hosts\u0027,"},{"line_number":330,"context_line":"            _no_resolv,"},{"line_number":331,"context_line":"            \u0027--strict-order\u0027,"},{"line_number":332,"context_line":"            \u0027--except-interface\u003dlo\u0027,"},{"line_number":333,"context_line":"            \u0027--pid-file\u003d%s\u0027 % pid_file,"},{"line_number":334,"context_line":"            \u0027--dhcp-hostsfile\u003d%s\u0027 % self.get_conf_file_name(\u0027host\u0027),"}],"source_content_type":"text/x-python","patch_set":1,"id":"5f93b717_3b5ac54e","side":"PARENT","line":331,"updated":"2018-01-29 16:28:29.000000000","message":"Changing this is a double-edged sword, and old comment I know, https://www.redhat.com/archives/libvir-list/2007-May/msg00005.html :\n\n\"The list of nameservers is a prioritized lists, so by trying a random \nnameserver you may get different DNS results returned, compared to using \nthem in sorted orders. Nameservers may also be ordered by locality, for\nexample the first 2 nameservers are on my local LAN, but the 3rd nameserver\nis a \u0027failsafe\u0027 on the WAN. The first will give much faster loookup result\nthat the last.\"\n\nBut then I also see people saying you should never use --strict-order with dnsmasq.\n\nI\u0027m not sure how we can make everyone happy, but also don\u0027t know why this is the default, just wanted to put the info out there.","commit_id":"03b1cca794592a4899ccb0a16474f9681f203e29"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"eaaaf9ac04b0a737e3275b90085d8354a99da07a","unresolved":false,"context_lines":[{"line_number":328,"context_line":"            \u0027dnsmasq\u0027,"},{"line_number":329,"context_line":"            \u0027--no-hosts\u0027,"},{"line_number":330,"context_line":"            _no_resolv,"},{"line_number":331,"context_line":"            \u0027--strict-order\u0027,"},{"line_number":332,"context_line":"            \u0027--except-interface\u003dlo\u0027,"},{"line_number":333,"context_line":"            \u0027--pid-file\u003d%s\u0027 % pid_file,"},{"line_number":334,"context_line":"            \u0027--dhcp-hostsfile\u003d%s\u0027 % self.get_conf_file_name(\u0027host\u0027),"}],"source_content_type":"text/x-python","patch_set":1,"id":"5f93b717_f006b786","side":"PARENT","line":331,"in_reply_to":"5f93b717_3b5ac54e","updated":"2018-01-29 18:59:12.000000000","message":"@Brian: good insight. I tend to agree that \u0027randomizing\u0027 the order of the queries may end up backfiring for other reasons. Frankly I am surprised that dnsmasq doesn\u0027t try other DNS servers it knows about when receiving SERVFAIL, but it seems this type of issue was reported before [1], and it looks to be the intended behavior from dnsmasq (comments [4,5] to name a few).\n\n[1] is very similar to what is being tackled here, and when looking at --all-servers option where it was proposed (right next to --strict-order in the man page [2]), it looks like it can mitigate the observed behavior. One might think this might have adverse performance effects, but caching that offset some of that[3].\n\n@Dirk: have you looked at that and discarded it? Can you elaborate? If not, I\u0027d suggest to give it a try instead of removing this option.\n\n [1] https://bugzilla.redhat.com/show_bug.cgi?id\u003d1400909\n [2] http://www.thekelleys.org.uk/dnsmasq/docs/dnsmasq-man.html\n [3] https://ma.ttwagner.com/make-dns-fly-with-dnsmasq-all-servers/\n [4] https://www.mail-archive.com/dnsmasq-discuss@lists.thekelleys.org.uk/msg11047.html\n [5] https://www.mail-archive.com/dnsmasq-discuss@lists.thekelleys.org.uk/msg10867.html","commit_id":"03b1cca794592a4899ccb0a16474f9681f203e29"},{"author":{"_account_id":1131,"name":"Brian Haley","email":"haleyb.dev@gmail.com","username":"brian-haley"},"change_message_id":"f0b7ea2f2fcdcc0c3a7cf68a195ecc5ce80bd9b3","unresolved":false,"context_lines":[{"line_number":328,"context_line":"            \u0027dnsmasq\u0027,"},{"line_number":329,"context_line":"            \u0027--no-hosts\u0027,"},{"line_number":330,"context_line":"            _no_resolv,"},{"line_number":331,"context_line":"            \u0027--strict-order\u0027,"},{"line_number":332,"context_line":"            \u0027--except-interface\u003dlo\u0027,"},{"line_number":333,"context_line":"            \u0027--pid-file\u003d%s\u0027 % pid_file,"},{"line_number":334,"context_line":"            \u0027--dhcp-hostsfile\u003d%s\u0027 % self.get_conf_file_name(\u0027host\u0027),"}],"source_content_type":"text/x-python","patch_set":1,"id":"5f93b717_8d3832a0","side":"PARENT","line":331,"in_reply_to":"5f93b717_451f7e5a","updated":"2018-01-31 21:56:19.000000000","message":"@Armando, thanks for looking further.  After reading that I would agree it\u0027s probably best to remove it.  Having it working, even if some responses are slow, is better than half-working, and a better user experience.","commit_id":"03b1cca794592a4899ccb0a16474f9681f203e29"},{"author":{"_account_id":748,"name":"Armando Migliaccio","email":"armamig@gmail.com","username":"armando-migliaccio"},"change_message_id":"8f86c16849a7c321ab1c9f915f0957a6c5322e02","unresolved":false,"context_lines":[{"line_number":328,"context_line":"            \u0027dnsmasq\u0027,"},{"line_number":329,"context_line":"            \u0027--no-hosts\u0027,"},{"line_number":330,"context_line":"            _no_resolv,"},{"line_number":331,"context_line":"            \u0027--strict-order\u0027,"},{"line_number":332,"context_line":"            \u0027--except-interface\u003dlo\u0027,"},{"line_number":333,"context_line":"            \u0027--pid-file\u003d%s\u0027 % pid_file,"},{"line_number":334,"context_line":"            \u0027--dhcp-hostsfile\u003d%s\u0027 % self.get_conf_file_name(\u0027host\u0027),"}],"source_content_type":"text/x-python","patch_set":1,"id":"5f93b717_451f7e5a","side":"PARENT","line":331,"in_reply_to":"5f93b717_f006b786","updated":"2018-01-31 04:37:18.000000000","message":"@Brian: I have given your insight a bit more thought and whilst I appreciate the use case as you present it, it\u0027s my understanding that the use of the strict-order doesn\u0027t yield the desired behavior (which as you described it would be great ;)). If the server chosen is always the last, that kinda defeats the purpose of having a list in the first place. On that basis, I\u0027d be tempted to say that removing the option would indeed be sensible as we would improve the availability of the DNS substrate by randomly using the resolvers. that backed by the statement made by the dnsmasq author himself [1], and I quote him (hopefully not out of context): \"That\u0027s one reason why \u0027strict-order\u0027 is broken and not recommended\", makes me wonder if Dirk is onto something here...\n\n[1] http://lists.thekelleys.org.uk/pipermail/dnsmasq-discuss/2009q3/003295.html","commit_id":"03b1cca794592a4899ccb0a16474f9681f203e29"}]}
