Re: Adding a new agent terminates existing executors?

Understood. Thanks for the help. On Wed, Nov 15, 2017 at 3:04 PM, Vinod Kone wrote: > Yes, there are a bunch of flags that need to be different. There are > likely some isolators which will not work correctly when you have multiple > agents on the same host even then. The

Re: Adding a new agent terminates existing executors?

Yes, there are a bunch of flags that need to be different. There are likely some isolators which will not work correctly when you have multiple agents on the same host even then. The garbage collector assumes it has sole access to the disk containing work dir etc etc. In general, running multiple

Re: Adding a new agent terminates existing executors?

Bingo. It probably doesn't hurt to differentiate --runtime_dir per agent but the real problem is that --cgroups_root needs to be different too. As one might infer from linux_launcher.cpp: Future LinuxLauncherProcess::recover( > const list& states) > { > // Recover all of the

Re: Adding a new agent terminates existing executors?

> On Nov 15, 2017, at 8:24 AM, Dan Leary wrote: > > Yes, as I said at the outset, the agents are on the same host, with different > ip's and hostname's and work_dir's. > If having separate work_dirs is not sufficient to keep containers separated > by agent, what

Re: Adding a new agent terminates existing executors?

Yes, as I said at the outset, the agents are on the same host, with different ip's and hostname's and work_dir's. If having separate work_dirs is not sufficient to keep containers separated by agent, what additionally is required? On Wed, Nov 15, 2017 at 11:13 AM, Vinod Kone

Re: Adding a new agent terminates existing executors?

How is agent2 able to see agent1's containers? Are they running on the same box!? Are they somehow sharing the filesystem? If yes, that's not supported. On Wed, Nov 15, 2017 at 8:07 AM, Dan Leary wrote: > Sure, master log and agent logs are attached. > > Synopsis: In the

Re: Adding a new agent terminates existing executors?

That seems weird then. A new agent coming up on a new ip and host, shouldn't affect other agents running on different hosts. Can you share master logs that surface the issue? On Tue, Nov 14, 2017 at 12:51 PM, Dan Leary wrote: > Just one mesos-master (no zookeeper) with

Re: Adding a new agent terminates existing executors?

Just one mesos-master (no zookeeper) with --ip=127.0.0.1 --hostname=localhost. In /etc/hosts are 127.1.1.1agent1 127.1.1.2agent2 etc. and mesos-agent gets passed --ip=127.1.1.1 --hostname=agent1 etc. On Tue, Nov 14, 2017 at 3:41 PM, Vinod Kone wrote: >

Adding a new agent terminates existing executors?

So I have a bespoke framework that runs under 1.4.0 using the v1 HTTP API, custom executor, checkpointing disabled. When the framework is running happily and a new agent is added to the cluster all the existing executors immediately get terminated. The scheduler is told of the lost executors and