Hi Folks,
Have been using docker and docker-compose inside Virtualbox VMs running
Alpine 3.13 guestOS, on a Windows 10 Professional host. The Alpine Guest
has 2 vCPUs, 4GB vRAM, and even when all containers are running the RAM
usage remains below 1GB.
- Docker version 20.10.3, build 48d30b5b32e99c932b4ea3edca74353feddd83ff
- docker-compose version 1.27.4, build unknown
I use 2 docker-compose .yml files to bring up, 2 set of containers. First
.yml file brings up Telgraf, InfluxDB, Grafana containers based on images
from DockerHub. The second .yml brings up a POJO-server (Java) with
application and another 3 tester/simulators.They all share a dedicated
network ‘mynet’ (using bridged driver). The above setup seems to work
perfectly at times. For example:
-
Sometimes , containers don’t seem to get hostname assigned (i.e. the
mnemonic container-id is set as hostname, as is default) while at other
times specific hostname is indeed assigned.
-
Then, for a few containers (not all), the inter-container DNS/name
resolution doesn’t seem to work, but at the same time few other containers
have perfectly working inter-container DNS/name resolution. When
inter-container DNS resolution does-not, if I log into the container
runtime using ‘docker exec -it bash’, and try to do 'ping ', I get the
following error:
*ping: somehostname: Temporary failure in name resolution*
-
Sometimes, some containers (not all) require the "service: > hostname: "
declaration in the docker-compose .yml file to get specific hostname
assigned to the container, but for other containers at other times, that
declaration is not needed, and only the “container_name” declaration is
enough (in .yml file) to have specific hostname for that container !!
Wondering if anyone can help figure out why is the behaviour erratic ? What
can I do to fix it ? Is this an issue with Docker only on Alpine ?
Thanks in anticipation.
regards,
BD
I just ran into this as well. I’m not using it for production so I wasn’t worried about it until I saw this email. Restarting the docker daemon seemed to fix things. I hadn’t used my Docker system for a almost year (working on other projects) and didn’t see this problem back then.
My docker system is:
Alpine 3.12.4
Docker version 20.10.3, build 48d30b5b32e99c932b4ea3edca74353feddd83ff
docker-compose version 1.25.4, build unknown
Running as a qemu guest on Proxmox 6.3-4, 1.11 GB RAM used of 16 GB allocated, 4 vCPUs available. 570 MB used out of a 16 GB boot disk. Docker images and containers are stored on an iSCSI mount coming from a FreeNAS server hosted on the same hardware.
Underlying hardware is a SuperMicro X11SDV-4C-TLN2F Server Motherboard - Intel Xeon D-2123IT Quad-core (4 Core) 2.20 GHz, 64 GB RAM, LSI 9207-8i HBA, 8 x 4 GB Ironwolf drives, 3 x Samsung 860 EVO 250 GB SSD, 1 x Samsung EVO 860 500 GB SSD.
Is it worth filing a bug? And where should the bug be filed?
—Paul
> > On Mar 3, 2021, at 4:11 PM, Banibrata Dutta <banibrata.dutta@gmail.com> wrote:> > Hi Folks,> > Have been using docker and docker-compose inside Virtualbox VMs running Alpine 3.13 guestOS, on a Windows 10 Professional host. The Alpine Guest has 2 vCPUs, 4GB vRAM, and even when all containers are running the RAM usage remains below 1GB.> > • Docker version 20.10.3, build 48d30b5b32e99c932b4ea3edca74353feddd83ff> • docker-compose version 1.27.4, build unknown> I use 2 docker-compose .yml files to bring up, 2 set of containers. First .yml file brings up Telgraf, InfluxDB, Grafana containers based on images from DockerHub. The second .yml brings up a POJO-server (Java) with application and another 3 tester/simulators.They all share a dedicated network ‘mynet’ (using bridged driver). The above setup seems to work perfectly at times. For example:> > • Sometimes , containers don’t seem to get hostname assigned (i.e. the mnemonic container-id is set as hostname, as is default) while at other times specific hostname is indeed assigned.> > • Then, for a few containers (not all), the inter-container DNS/name resolution doesn’t seem to work, but at the same time few other containers have perfectly working inter-container DNS/name resolution. When inter-container DNS resolution does-not, if I log into the container runtime using ‘docker exec -it bash’, and try to do 'ping ', I get the following error:> ping: somehostname: Temporary failure in name resolution> > • Sometimes, some containers (not all) require the "service: > hostname: " declaration in the docker-compose .yml file to get specific hostname assigned to the container, but for other containers at other times, that declaration is not needed, and only the “container_name” declaration is enough (in .yml file) to have specific hostname for that container !!> > Wondering if anyone can help figure out why is the behaviour erratic ? What can I do to fix it ? Is this an issue with Docker only on Alpine ?> > Thanks in anticipation.> > regards,> BD
On Thu, Mar 4, 2021 at 6:56 AM Paul Suh <plsuh@goodeast.com> wrote:
> I just ran into this as well. I’m not using it for production so I wasn’t> worried about it until I saw this email. Restarting the docker daemon> seemed to fix things. I hadn’t used my Docker system for a almost year> (working on other projects) and didn’t see this problem back then.>> My docker system is:> Alpine 3.12.4> Docker version 20.10.3, build 48d30b5b32e99c932b4ea3edca74353feddd83ff> docker-compose version 1.25.4, build unknown>> Running as a qemu guest on Proxmox 6.3-4, 1.11 GB RAM used of 16 GB> allocated, 4 vCPUs available. 570 MB used out of a 16 GB boot disk. Docker> images and containers are stored on an iSCSI mount coming from a FreeNAS> server hosted on the same hardware.>> Underlying hardware is a SuperMicro X11SDV-4C-TLN2F Server Motherboard -> Intel Xeon D-2123IT Quad-core (4 Core) 2.20 GHz, 64 GB RAM, LSI 9207-8i> HBA, 8 x 4 GB Ironwolf drives, 3 x Samsung 860 EVO 250 GB SSD, 1 x Samsung> EVO 860 500 GB SSD.>
Thanks for the reply. In my case I've tried several things, which do not
help consistently:
- Restarting docker daemon
- Restarting the VM
- Ensuring the Virtualbox SW is latest on the host system
Only thing that I'm yet to do is to reinstall docker and docker-compose. I
was a bit wary of doing that, but might as well try it.
Is it worth filing a bug? And where should the bug be filed?
>
Possibly. If our issues are indeed similar, either of us can create a bug
reporrt, and other person can upvote it. Need to figure out how to report
bugs on Alpine.
> —Paul>>> >> > On Mar 3, 2021, at 4:11 PM, Banibrata Dutta <banibrata.dutta@gmail.com>> wrote:> >> > Hi Folks,> >> > Have been using docker and docker-compose inside Virtualbox VMs running> Alpine 3.13 guestOS, on a Windows 10 Professional host. The Alpine Guest> has 2 vCPUs, 4GB vRAM, and even when all containers are running the RAM> usage remains below 1GB.> >> > • Docker version 20.10.3, build> 48d30b5b32e99c932b4ea3edca74353feddd83ff> > • docker-compose version 1.27.4, build unknown> > I use 2 docker-compose .yml files to bring up, 2 set of containers.> First .yml file brings up Telgraf, InfluxDB, Grafana containers based on> images from DockerHub. The second .yml brings up a POJO-server (Java) with> application and another 3 tester/simulators.They all share a dedicated> network ‘mynet’ (using bridged driver). The above setup seems to work> perfectly at times. For example:> >> > • Sometimes , containers don’t seem to get hostname assigned (i.e.> the mnemonic container-id is set as hostname, as is default) while at other> times specific hostname is indeed assigned.> >> > • Then, for a few containers (not all), the inter-container> DNS/name resolution doesn’t seem to work, but at the same time few other> containers have perfectly working inter-container DNS/name resolution. When> inter-container DNS resolution does-not, if I log into the container> runtime using ‘docker exec -it bash’, and try to do 'ping ', I get the> following error:> > ping: somehostname: Temporary failure in name resolution> >> > • Sometimes, some containers (not all) require the "service: >> hostname: " declaration in the docker-compose .yml file to get specific> hostname assigned to the container, but for other containers at other> times, that declaration is not needed, and only the “container_name”> declaration is enough (in .yml file) to have specific hostname for that> container !!> >> > Wondering if anyone can help figure out why is the behaviour erratic ?> What can I do to fix it ? Is this an issue with Docker only on Alpine ?> >> > Thanks in anticipation.> >>>