search

containers / aardvark-dns

Authoritative dns server for A/AAAA container records. Forwards other request to host's /etc/resolv.conf
Apache License 2.0
187 stars 32 forks source link

dns request failed: request timed out #248

Open MartinX3 opened 2 years ago

MartinX3 commented 2 years ago

I'm using arch linux, so the packages should have the newest version.

I'm using firewallD and rootless podman with netavark and aardvark-dns.

I understand, that rootless podman with netavark won't manage my firewallD, but I would like to know which rules I need to activate to avoid the spam in my journal. And if the rule need to be in my loopback or network interface. (Also if it is enough to allow communication with the host instead of having an open port in the internet.

My dns resolver is systemd-resolved

$ ls -lha /etc/resolv.conf
lrwxrwxrwx 1 root root 39 31. Okt 10:22 /etc/resolv.conf -> ../run/systemd/resolve/stub-resolv.conf

My journal spam: aardvark-dns[6156]: 21433 dns request failed: request timed out

The rootless container itself can ping to google.com. I didn't test if they can ping to a container dns name.

MartinX3 commented 2 years ago

Ah, after turning /etc/resolv.conf again into a symlink of systemd-resolved sudo ln -rsf /run/systemd/resolve/stub-resolv.conf /etc/resolv.conf the messages disappear. I assume it wants now to call the localhost ip of the systemd-resolved instead of the network ip of the outside resolver.

I don't know how I could let this plugin verbose debug logging into the journal, but I'm glad that it is fixed now.

MartinX3 commented 2 years ago

And it's back dns request failed: request timed out

It seems to happen if I execute nslookup in a container. But I get a result. Weird.

Szwendacz99 commented 1 year ago

I detected the same problem on my raspberry pi with Fedora and rootless podman containers.

jmaris commented 1 year ago

Having the exact same issue. DNS lookup is extremely unstable, only two thirds of lookups work.

This is on Fedora Silverblue 37 using rootless containers and the new networking stack.

MartinX3 commented 1 year ago

Now using a nextcloud container which communicates with a PostgreSQL and a LDAP container. I access the nextcloud container with a nginx reverse proxy container from the internet.

1/3 of the nextcloud web interface results into a 502 bad gateway, because of the unstable DNS. And my logs are getting spammed with

aardvark-dns[3617]: 50310 dns request failed: request timed out

I hope the next podman release will include the new network stack which hopefully fixes this issue.

baude commented 1 year ago

It is impossible to help with these issues as reporters did not provide versions for podman, netavark, and aarvark-dns. please provide as much relevant information as possible.

MartinX3 commented 1 year ago

podman: 4.3.1 netavark: 1.4.0 aardvark-dns: 1.4.0

baude commented 1 year ago

would you say your machine/vm is high performent or maybe has slow IO/processor/RAM limitations? I'm trying to understand if you might have a race.

MartinX3 commented 1 year ago

The first bare metal server server with this issue has the HDD connected to SATA with the server. CPU: Intel Xeon E3-1231 v3 - 3.4 GHz - 4 core(s) RAM: 32GB - DDR3

The second bare metal server has a 870 Evo SSD connected via SATA to the server. CPU: AMD Phenom 2 955 X4 3.2 GHz 4 cores RAM: 6GB - DDR2

The second server spams this issue many times. The first server spams it less often but also regularly.

baude commented 1 year ago

@flouthoc wdyt?

MartinX3 commented 1 year ago

Now my faster server spammed it at night while the server pods weren't used by clients.

flouthoc commented 1 year ago

We used to see similar issues in older versions of netavark and aardvark in Podman CI as well but it was fixed in newer versions with https://github.com/containers/aardvark-dns/pull/220 but I guess there might be some issue which is not being reproduced in our CI, I'll try to reproduce this locally and see if i can reproduce this.

vulpes2 commented 1 year ago

I'm seeing the same problems on a freshly installed Fedora Server 37 instance as well. The machine is basically completely idling with no load and my journal is still filled with these errors. This issue was not present before 2022, and I've ran similar setups on slower machines without any DNS lookup issues.

podman: 4.3.1 netavark: 1.4.0 aardvark-dns: 1.4.0

Luap99 commented 1 year ago

Can you test with v1.5?

MartinX3 commented 1 year ago

I think the timeout is fixed. But now sometimes I get an "empty dns response" on all machines.

Luap99 commented 1 year ago

Does this cause problems for the container or is just an error that is logged often?

MartinX3 commented 1 year ago

It's logged often with long breaks between. So it appears in groups most of the time.

I think the services just repeat the DNS action again until it works. So I would say it just consumes CPU time and maybe network speed?

Maybe I don't use it long enough to see long term errors.

Luap99 commented 1 year ago

Do you have a simple reproducer? What kind of application are you running and how many dns request does it make?

vulpes2 commented 1 year ago

Tested with v1.5 and I'm getting a lot of dns request got empty response as well. Here's a list of all the containers I'm running on my system:

It's notable that none of these containers are particularly demanding on the hardware, and my system load average is generally below 0.1 at all times.

MartinX3 commented 1 year ago

It happens without workload.

The server just runs

soiboi commented 1 year ago

I'm experiencing the same issue with aardvark-dns reporting lots of dns request got empty response errors in my logs. It seems to be causing problems for at least the containers running Uptime Kuma, Invidious and Jellyseerr. Uptime Kuma starts throwing ECONNRESET when doing GET requests, and Invidious and Jellyseerr similarly start to have their requests fail, with external content taking a long time to load, if at all.

It happens with both 1.5.0 and the latest 1.6.0 from podman-next. For me it seems to start after around 3 days of uptime. I've tried changing machines and switching from onboard Realtek to an Intel i350-T2 controller, but both to no avail. Rebooting solves the issue, until uptime reaches 3 days again.

urbenlegend commented 8 months ago

Just ran into this issue as well with Nextcloud + Nginx Proxy Manager. What's funny is that I am using the same docker-compose setup on two different servers and one works fine while the other one doesn't. The only difference is that the one that is breaking isn't publicly accessible on the Internet and is instead setup to respond over a .lan domain which is configured on the home router. NPM has a proxy host setup that responds to mydomain.lan and redirects it to the nextcloud container.

It will work for a bit when I up/down NPM, but then eventually fail after a few hours or even days with 502 bad gateway errors, and dns request got empty response starts getting spammed into journalctl.

My setup

Here are my docker-compose files to set up each of them (rootful btw): Nextcloud docker-compose.yml

   version: '3'

   services:
     db:
       image: mariadb
       command: --transaction-isolation=READ-COMMITTED --log-bin=binlog --binlog-format=ROW
       restart: always
       volumes:
         - ./db:/var/lib/mysql
       environment:
         - MYSQL_ROOT_PASSWORD=<pw here>
         - MARIADB_AUTO_UPGRADE=1
         - MARIADB_DISABLE_UPGRADE_BACKUP=1
       env_file:
         - db.env
       networks:
         - backend

     redis:
       image: redis:alpine
       restart: always
       networks:
         - backend

     nextcloud:
       image: nextcloud:apache
       restart: always
       volumes:
         - ./html:/var/www/html
       environment:
         - MYSQL_HOST=db
         - REDIS_HOST=redis
       env_file:
         - db.env
       depends_on:
         - db
         - redis
       networks:
         - nextcloud_frontend
         - backend

     cron:
       image: nextcloud:apache
       restart: always
       volumes:
         - ./html:/var/www/html
       entrypoint: /cron.sh
       depends_on:
         - db
         - redis
       networks:
         - backend

   networks:
     nextcloud_frontend:
       external: true
     backend:

db.env

   MYSQL_PASSWORD=<pw here>
   MYSQL_DATABASE=nextcloud
   MYSQL_USER=nextcloud

Nginx Proxy Manager docker-compose.yml

   version: '3.8'
   services:
     proxy:
       image: 'jc21/nginx-proxy-manager:latest'
       restart: always
       ports:
         # These ports are in format <host-port>:<container-port>
         - '80:80' # Public HTTP Port
         - '443:443' # Public HTTPS Port
         - '81:81' # Admin Web Port
         # Add any other Stream port you want to expose
         # - '21:21' # FTP

       # Uncomment the next line if you uncomment anything in the section
       # environment:
         # Uncomment this if you want to change the location of
         # the SQLite DB file within the container
         # DB_SQLITE_FILE: "/data/database.sqlite"

         # Uncomment this if IPv6 is not enabled on your host
         # DISABLE_IPV6: 'true'

       healthcheck:
         test: ["CMD", "/bin/check-health"]
         interval: 30s
         timeout: 3s

       volumes:
         - ./data:/data
         - ./letsencrypt:/etc/letsencrypt

       networks:
         - nextcloud_frontend

   networks:
     nextcloud_frontend:
       external: true
Luap99 commented 1 month ago

Can anyone check if they still see this with aardvark-dns v1.12.2?

MartinX3 commented 1 month ago

@Luap99 At least for me it looks fine. Thank you.

lovette commented 3 weeks ago

I started getting a fair number of dns request got empty response from aardvark-dns out of the blue yesterday. I'm running podman from COPR and daily updates with dnf-automatic so I can't say for sure what version it started with. My system has been stable otherwise.

podman version 5.3.0-dev-29eb8ce09 aardvark-dns 1.13.0-dev

olegstepura commented 3 weeks ago

I encountered the same issue, and it turned out to be self-inflicted. I had added the podman network gateway IP to my /etc/resolv.conf, which I believe caused an infinite DNS resolution loop since Aardvark DNS might be using /etc/resolv.conf for upstream. Once I removed the podman network gateway IP from /etc/resolv.conf, the issue was resolved.

lovette commented 3 weeks ago

My resolv.conf seems to be in order.

# Generated by NetworkManager
nameserver 1.1.1.1
nameserver 1.0.0.1
nameserver 185.12.64.1
# NOTE: the libc resolver may not support more than 3 nameservers.
# The nameservers listed below may not be recognized.
nameserver 185.12.64.2
nameserver 2a01:4ff:ff00::add:2
nameserver 2a01:4ff:ff00::add:1

Over the course of a few hours yesterday (21:31 to 23:00 UTC), a bunch of empty response messages were logged, but none have been logged since.

Also network related, and also odd, is pasta has recently started logging No external routable interface for IPv6 on my dev machine.

podman run --rm -i --log-driver=passthrough-tty -v /codebase/webapp:/webapp dd2to3 pyup_dirs --recursive --py313-plus /webapp

Oct 16 13:30:53 dd-owen-dev pasta[79493]: No external routable interface for IPv6

% ip addr show
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
    valid_lft forever preferred_lft forever
    inet6 ::1/128 scope host 
    valid_lft forever preferred_lft forever
2: ens160: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether 00:0c:29:67:79:43 brd ff:ff:ff:ff:ff:ff
    altname enp3s0
    inet 172.16.209.6/24 brd 172.16.209.255 scope global noprefixroute ens160
    valid_lft forever preferred_lft forever
15: podman1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue state UP group default qlen 1000
    link/ether fe:39:d9:c2:2b:8a brd ff:ff:ff:ff:ff:ff
    inet 10.89.0.1/24 brd 10.89.0.255 scope global podman1
    valid_lft forever preferred_lft forever
    inet6 fe80::fc39:d9ff:fec2:2b8a/64 scope link 
    valid_lft forever preferred_lft forever
16: veth0@if2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master podman1 state UP group default qlen 1000
    link/ether e2:4f:e8:89:28:c4 brd ff:ff:ff:ff:ff:ff link-netns netns-9f79fe86-d8c5-bd93-4fef-0f95d4ec782c
    inet6 fe80::e04f:e8ff:fe89:28c4/64 scope link 
    valid_lft forever preferred_lft forever
17: veth1@if2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master podman1 state UP group default qlen 1000
    link/ether b2:12:13:3e:d7:d1 brd ff:ff:ff:ff:ff:ff link-netns netns-264e8cb4-a3a2-84ec-4dad-5d146bff3f72
    inet6 fe80::b012:13ff:fe3e:d7d1/64 scope link 
    valid_lft forever preferred_lft forever
18: veth2@if2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master podman1 state UP group default qlen 1000
    link/ether fe:fa:ac:70:ca:85 brd ff:ff:ff:ff:ff:ff link-netns netns-10327a44-fdc9-099e-340f-17ddfd759e0d
    inet6 fe80::fcfa:acff:fe70:ca85/64 scope link 
    valid_lft forever preferred_lft forever
19: veth3@if2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master podman1 state UP group default qlen 1000
    link/ether 1e:20:af:e6:df:23 brd ff:ff:ff:ff:ff:ff link-netns netns-e70477fb-f6e9-ade7-c624-2b4ce47053be
    inet6 fe80::1c20:afff:fee6:df23/64 scope link 
    valid_lft forever preferred_lft forever
20: veth4@if2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master podman1 state UP group default qlen 1000
    link/ether d2:3f:47:c2:e3:91 brd ff:ff:ff:ff:ff:ff link-netns netns-448fb15e-a452-efde-7ae0-87d925f94091
    inet6 fe80::d03f:47ff:fec2:e391/64 scope link 
    valid_lft forever preferred_lft forever
21: veth5@if2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master podman1 state UP group default qlen 1000
    link/ether 9e:91:56:22:b6:b3 brd ff:ff:ff:ff:ff:ff link-netns netns-6e0e3dfe-5214-81d4-ea80-043e5844a1e4
    inet6 fe80::9c91:56ff:fe22:b6b3/64 scope link 
    valid_lft forever preferred_lft forever
22: veth6@if2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master podman1 state UP group default qlen 1000
    link/ether fa:cf:0b:9c:28:c7 brd ff:ff:ff:ff:ff:ff link-netns netns-c3bfdebf-2e88-8d7d-d3cd-2060a8c7428a
    inet6 fe80::f8cf:bff:fe9c:28c7/64 scope link 
    valid_lft forever preferred_lft forever
23: veth7@if2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master podman1 state UP group default qlen 1000
    link/ether 66:71:f8:9e:67:bc brd ff:ff:ff:ff:ff:ff link-netns netns-c72a6ec6-a03e-579c-8d1f-86a6719d86b7
    inet6 fe80::6471:f8ff:fe9e:67bc/64 scope link 
    valid_lft forever preferred_lft forever
24: veth8@if2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master podman1 state UP group default qlen 1000
    link/ether 4a:3a:76:80:59:9b brd ff:ff:ff:ff:ff:ff link-netns netns-76238717-eeb9-50a5-f96e-3aac5e4e7d31
    inet6 fe80::483a:76ff:fe80:599b/64 scope link 
    valid_lft forever preferred_lft forever
25: veth9@if2: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc noqueue master podman1 state UP group default qlen 1000
    link/ether c2:c6:30:88:ad:56 brd ff:ff:ff:ff:ff:ff link-netns netns-41c7d6c5-76f6-11f3-f772-06f999175ed6
    inet6 fe80::c0c6:30ff:fe88:ad56/64 scope link 
    valid_lft forever preferred_lft forever

Other than dnf-automatic package installs, not much about my dev machine has changed in the better part of a year.