Here Mesosphere recommends running Mesos-DNS through Marathon on one of the slaves. However, the nameservers on the DNS clients are configured statically. So when a node fails and the Mesos-DNS task gets scheduled on an other node, the nameserver in resolv.conf won't be correct anymore. What's the idea behind this, and how is this usually solved?
Yea, something like that. I'd suggest the first 2 nameservers in /etc/resolv.conf be mesos-dns servers, and the 3rd be a non mesos-dns. The reason is by default, only the first 3 nameserver entries are looked at (
man resolv.conf, search for MAXNS). If you ever bring down mesos-dns you'd at least have a third entry in /etc/resolv.conf that can resolve queries.
Example. My internal top-level domain is .galactic-empire domain. I have a master/slave node named sidious and a slave node named eclipse. leader.galactic-empire resolves to the sidious ip, slave.galactic-empire resolves both the sidious and eclipse IPs.
dns.marathon.galactic-empire resolves correctly to point at sidious.
However, I can't seem to resolve the sidious or eclipse hostnames.
I would have expected eclipse.slave.galactic-empire to resolve or sidious.leader.galactic-empire to resolve.
but they do not.
It seems as if https://github.com/mesosphere/mesos-dns/blob/master/records/generator.go#L241 answers my question.
Do any devs actually look at this chat?
Hi, I'm testing out mesos + marathon and zookeeper in the simplest setup of a single master and a single slave node, by following the instructions from https://open.mesosphere.com/getting-started/install/. Both the master and the slave are ubuntu 14.04 LTS (trusty) instances on a private open stack cloud, and they both communicate through password-authenticated ssh. while installation went through, I wasn't able to start the mesos service properly. Marathon runs OK. I suspect this is all due to the fact that I am not establishing the proper communication channels between master and slave via ssh.... In particular, when I run: sudo mesos-slave --master=username:password@masterIP:2888/etc/zookeeper, I get the following error: ZOO_ERROR@handle_socket_error_msg@1643... connection to masterIP:2888 timed out. Does anyone know what I am missing in my installation / configuration? Any help is greatly appreciated.
Hi, just a small hint: in order to receive answers on your question regarding Mesos-DNS you can also try to ask them in the Mesosphere community Slack channel : http://chat.mesosphere.com/
I m trying to link using container hostname/name by setting key value pair in marathon json. but its not connecting to other contianer
@muraliraju2013 for that you require service discovery, in order for your services to discover and find out where other services are running, you must not pass the host name to marathon json. as the services might be re-scheduled at any point of time and run on different mesos-slave host.
checkout these systems: consul, mesos-dns, marathon-dns,
@alaa I'm not getting the full picture here. Are you saying like rather than using the linking option, I should mention the hostname and port explicitly?
As you see, in my case the application needs to communicate with the database. Should I mention the hostname within the db configuration of the application? I saw that i can get the hostname using the Mesos-DNS service and I indeed see the HTTP APIs for that. But, i'm not yet able to figure the method.
As you see, in my case the application needs to communicate with the database. Should I mention the hostname within the db configuration of the application? I saw that i can get the hostname using the Mesos-DNS service and I indeed see the HTTP APIs for that. But, i'm not yet able to figure the method.
hello everyone. i started following instructions here: https://open.mesosphere.com/advanced-course/building-and-running-mesos-dns/ and im on exercise 7. The godep build process is running for about 2 hours now - is that normal?
hey everyone, a question that I got when using mesos with a LB (used marathon LB):
currently, the xxx.marathon.mesos hostname will resolve to the IP of the node (host) in which the container is running
when using the marathon LB, if I want to resolve the same hostname related to a Task (xxx.marathon.mesos) to the marathon LB
how can I do it?
currently, the xxx.marathon.mesos hostname will resolve to the IP of the node (host) in which the container is running
when using the marathon LB, if I want to resolve the same hostname related to a Task (xxx.marathon.mesos) to the marathon LB
how can I do it?
basically the communication should happen via the LB, not directly from host to host
Hello everyone. I started following this tutorial: https://open.mesosphere.com/advanced-course/building-and-running-mesos-dns/ on exercise 7.
After the command godep go build the following message appears:
'#'github.com/mesosphere/mesos-dns/httpcli
httpcli/doer.go:72: invalid operation: token.Claims["uid"] (type jwt.Claims does not support indexing)
httpcli/doer.go:73: invalid operation: token.Claims["exp"] (type jwt.Claims does not support indexing)
godep: go exit status 2
Someone can help me?
Thanks
After the command godep go build the following message appears:
'#'github.com/mesosphere/mesos-dns/httpcli
httpcli/doer.go:72: invalid operation: token.Claims["uid"] (type jwt.Claims does not support indexing)
httpcli/doer.go:73: invalid operation: token.Claims["exp"] (type jwt.Claims does not support indexing)
godep: go exit status 2
Someone can help me?
Thanks
does mesos-dns poll master or it has some mechanism to receive notifications from master?
knoldus@knoldus-Vostro-3558:~/dcos-vagrant$ vagrant up
Validating Plugins...
Validating User Config...
Validating Installer Checksum...
Using DC/OS Installer: installers/dcos/dcos_generate_config-1.8.8.sh
Using DC/OS Config: etc/config-1.8.yaml
Validating Machine Config...
Configuring VirtualBox Host-Only Network...
>>> Creating VirtualBox Network
0%...
Progress state: NS_ERROR_FAILURE
VBoxManage: error: Failed to create the host-only adapter
VBoxManage: error: VBoxNetAdpCtl: Error while adding new interface: failed to open /dev/vboxnetctl: No such file or directory
VBoxManage: error: Details: code NS_ERROR_FAILURE (0x80004005), component HostNetworkInterfaceWrap, interface IHostNetworkInterface
VBoxManage: error: Context: "RTEXITCODE handleCreate(HandlerArg*)" at line 71 of file VBoxManageHostonly.cpp
VirtualBox is complaining that the installation is incomplete. Please
run `VBoxManage --version` to see the error message which should contain
instructions on how to fix this error.
knoldus@knoldus-Vostro-3558:~/dcos-vagrant$ VBoxManage --version
WARNING: The character device /dev/vboxdrv does not exist.
Please install the virtualbox-dkms package and the appropriate
headers, most likely linux-headers-generic.
You will not be able to start VMs until this problem is fixed.
5.1.6_Ubuntur110634
knoldus@knoldus-Vostro-3558:~/dcos-vagrant$
I have tried every link and reinstalled the virtual box but same error persists
Hi @piyushknoldus, I don't think you'll receive an answer here any soon. Please try to use the community channels (Google User Groups and Slack) listed here: https://dcos.io/community/
@piyushknoldus, did you try this link: http://askubuntu.com/questions/787471/virtualbox-5-is-unable-to-create-a-host-only-adapter-on-ubuntu-16-04
hi all I m trying to run the Mesos DC/OS cluster using vagrant installation but is facing the error everytime I start it . Error : - A VirtualBox machine with the name 'm1.dcos' already exists. Please use another name or delete the machine with the existing name, and try again
I have also raised the question on stackoverflow with all mine finding and step , please have a look and help : https://stackoverflow.com/questions/42806665/a-virtualbox-machine-with-the-name-m1-dcos-already-exists-please-use-another
@adyatlov yes I tried that , i m able to installed it successfull but now the above error persists
can you give some attention to the issue where mesos-dns is not detecting ALL networked ips