| fredl | Good morning folks | 07:27 |
|---|---|---|
| fredl | So about 16.04 servers which have been upgraded from 16.04 to 18.04... I learned yesterday that this means that netplan is usually not set up and the old /etc/network/interfaces is still used unless other actians are taken | 07:28 |
| lordievader | Good morning | 07:43 |
| fredl | I'm trying to find a way where for a certain domain only one DNS will be used, if that DNS is down somehow it should not try to resolve using other resolvers | 08:14 |
| fredl | Somebody pointed met to the networkd config yesterday instead of systemd-resolvd | 08:14 |
| lordievader | Is this from the authoritative side or from the client side? | 08:26 |
| fredl | client | 08:55 |
| fredl | Situation is that we have a network gateway with a VPN tunnel to Google cloud | 08:55 |
| fredl | Client have a mere route for the GCP services to that gateway | 08:56 |
| fredl | So I'm trying to figure out how I'd configure networkd to 1) for domain such.com go to DNS gcp ONLY, 2) for other domains use 'normal' DNS servers. In any case we need to prevent that queries for such.com NEVER go to the default DNS servers, also in case of VPN failure or so | 08:57 |
| lordievader | In that case setting up a single resolver should do what you want. | 08:58 |
| fredl | I tried first to simply add the Google Cloud DNS server to the list of resolvers | 08:58 |
| fredl | But then the customer simulated a down VPN by blocking port 53 | 08:59 |
| fredl | Apparently that results in long, unwanted negative caching as the default DNS server can't resolve those queries then | 08:59 |
| lordievader | It should just give timeouts (which can be long), but not negative caching. | 09:04 |
| fredl | Correct | 09:13 |
| fredl | I configured systemd-resolved to use Cache=no-negative | 09:13 |
| fredl | So it shouldn't do negative caching at all anyway | 09:13 |
| fredl | in debugging output that's also what I see: Not caching negative entry for: blabla.nl IN AAAA, cache mode set to no-negative | 09:14 |
| fredl | So that's why I'm assuming that somehow our main DNS keeps a negative cache somehow? | 09:14 |
| fredl | And that after the VPN link goes down, perhaps systemd-resolved remembers the DNS it *got* that from and queries that one again and again | 09:15 |
| fredl | That's the only plausible explanation I see. | 09:16 |
| Odd_Bloke | rbasak: Congrats on git-ubuntu 1.0! | 14:49 |
| rbasak | Thanks! | 15:00 |
| tsukasadt | Hopefully an easy question: can you combine `grep` and `tail` to realtime track a log file but only get entries containing a specific string? If so, how? | 17:13 |
| sdeziel | tsukasadt: tail -f /var/log/mylog | grep 'my pattern of interest' | 17:13 |
| tsukasadt | Now I really feel stupid. I had almost that, but I put the file in the grep. ^^' | 17:15 |
| tsukasadt | ty, sdeziel | 17:15 |
| sdeziel | np | 17:15 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!