2024-10-21 GnuCash IRC logs
01:40:59 *** fell has quit IRC
01:42:18 *** fell has joined #gnucash
01:42:18 *** ChanServ sets mode: +o fell
02:13:59 <fell> warlord, jralls_afk: Weblate got 10h ago: https://www.gnucash.org/ (HTTPSConnectionPool(host='www.gnucash.org', port=443): Max retries exceeded with url: / (Caused by NameResolutionError("<urllib3.connection.HTTPSConnection object at 0x7f8376c7cb90>: Failed to resolve 'www.gnucash.org' ([Errno -3] Temporary failure in name resolution)")))
04:56:10 *** tmrsm has quit IRC
05:01:16 *** tmrsm has joined #gnucash
07:15:19 <warlord> fell, interesting. Can you get to www.gnucash.org ?
07:15:37 <warlord> (I can't test from here -- I have a local version of the domain)
07:16:19 <warlord> Actually,I CAN test it locally.. And I have a DNS failure, too...
07:16:30 <warlord> Oh, nevermind -- I DID get an answer..
07:16:38 <warlord> ;; ANSWER SECTION:
07:16:38 <warlord> www.gnucash.org. 86222 IN A 67.198.37.17
07:39:24 <Simon> two of the nameservers weren't responding when I checked earlier
07:39:41 *** lorimark has quit IRC
07:40:01 <Simon> they're still not responding (ns.linas.org [gnucash.org] (99.153.64.178) and ns.gnucash.org [gnucash.org] (99.153.64.179))
07:41:30 <Simon> this looks like a misconfiguration
07:41:47 <Simon> .org has been configured with the following glue:
07:41:49 <Simon> ns.linas.org. 3600 IN A 99.153.64.178
07:41:51 <Simon> ns.gnucash.org. 3600 IN A 99.153.64.179
07:41:53 <Simon> ns1.linas.org. 3600 IN A 67.198.37.16
07:41:55 <Simon> ns1.gnucash.org. 3600 IN A 67.198.37.17
07:42:11 <Simon> but the nameservers respond with different IPs:
07:42:13 <Simon> ns.linas.org. 86400 IN A 67.198.37.16
07:42:15 <Simon> ns.gnucash.org. 86222 IN A 67.198.37.17
07:42:17 <Simon> ns1.linas.org. 86400 IN A 67.198.37.16
07:42:19 <Simon> ns1.gnucash.org. 86222 IN A 67.198.37.17
07:49:09 <warlord> Does .org not know about fly-by-night?
07:52:01 <warlord> Simon, that glue DOES include the correct IPS for ns1 (67.198). I don't understand why it doesn't include fbn.
07:52:25 <warlord> fbn is in the whois response, so it should be in the glue.
07:52:47 <warlord> But it's possible that linas' servers don't respond properly.
07:53:49 <warlord> /I/ get responses for www.gnucash.org from both ns1.linas and ns1.gnucash
07:56:34 <Simon> it does know about fly-by-night, I just omitted those
07:56:46 <Simon> 99.153.64.178 and 99.153.64.179 don't respond to me
07:56:58 <Simon> if those are correct then the IPs in the zone are wrong
08:00:35 <warlord> I am pretty sure those are old addresses.
08:00:57 <Simon> you've just told me those are the correct IPs
08:01:14 <warlord> The 67.198 addresses are correct. The 99.153 are not.
08:01:17 <Simon> oh
08:01:22 <Simon> then the .org glue is wrong
08:01:56 <warlord> The org glue just has "extra data" -- it should not stop a properly-configured client from falling back to a correctly-working server.
08:02:05 <warlord> But yeah, Linas should probably remove the old glue.
08:20:37 <Simon> No, glue is not "extra data"
08:20:47 <Simon> It MUST be correct
08:21:25 <Simon> There are some resolvers that prefer parent data to child data that might not update it, ever
08:21:59 <Simon> So queries will tend to fail unless the domain is popular enough to cache that those IPs don't work
08:44:48 <warlord> Simon, I think you misunderstood what I meant by "extra data".. The glue records contain 5 servers. There are really only 3. Those 3 /are/ correctly in the glue. The only issue is that the glue contains 2 extra records (that are incorrect).
08:45:23 <warlord> A properly configured client should correctly fall back to another glue-record if one fails.
08:45:51 <warlord> (for the record, I was part of the IETF DNS Directorate for a decade; I am quite familiar with how DNS works.
09:17:24 <Simon> It takes time to timeout and retry other IPs which is what's causing resolution failures
09:20:22 <warlord> It should fall back after a few seconds.
09:20:33 <warlord> But yes, LInas should clean up the glue. I'll email him about it.
09:28:48 <warlord> Simon, I've notified Linas and asked him to clean it up.
09:59:38 <Simon> The effects are cumulative. By the time it has tried the two that don't respond the original client may have given up
10:14:17 *** afdee1c has joined #gnucash
11:22:57 <warlord> Fair. anyways, I did ask Linas to clean it up. Unfortunately he is the only one who can.
11:40:15 *** afdee1c has quit IRC
13:25:24 *** fell has quit IRC
13:25:43 *** fell has joined #gnucash
13:25:43 *** ChanServ sets mode: +o fell
14:07:02 *** afdee1c has joined #gnucash
14:46:16 <Simon> there's a discrepancy with fly-by-night.ihtfp.org too: https://dnsviz.net/d/gnucash.org/servers/
15:47:29 <fell> ^ warlord
15:48:32 <warlord> Simon, no way to avoid that -- FBN has multiple addresses, but I can only put one into the glue.
15:55:03 <Simon> fairly sure that's not true
15:55:46 <Simon> I can register multiple addresses for a .org nameserver
15:57:50 <warlord> Simon, my registrar does not allow it.
15:58:18 <warlord> Last I attempted to set it, they would only let me have 1 IPv4 and 1 IPv6 address in a glue record.
15:58:49 <warlord> My personal view is that, so long as the glue is /correct/, that is sufficient -- even if the glue is not /complete/
17:23:15 *** FeuerwehrMichael has joined #gnucash
17:23:15 *** ChanServ sets mode: +v FeuerwehrMichael
17:30:34 *** FeuerwehrMichael has quit IRC
18:46:29 *** afdee1c has quit IRC
19:21:18 *** warlord has quit IRC
19:28:17 *** afdee1c has joined #gnucash
21:01:33 *** warlord has joined #gnucash
21:01:33 *** gncbot sets mode: +o warlord
21:28:48 *** afdee1c has quit IRC
22:02:19 *** starman has quit IRC
22:02:26 *** starman has joined #gnucash