[ 
https://issues.apache.org/jira/browse/GEODE-1935?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15543050#comment-15543050
 ] 

Jens Deppe commented on GEODE-1935:
-----------------------------------

I've looked at this a bit and can get to a partial solution. When running a 
locator instance in a Docker container I first start the container and expose 
ports {{1099}} (default JMX port) and {{10334}} (default locator port). Then I 
start gfsh with this command:
{noformat}
gfsh start locator --name=locator1 \
  --J=-Dcom.sun.management.jmxremote.rmi.port=1099 \
  --J=-Djava.rmi.server.hostname=localhost \
  --J=-Dgemfire.jmx-manager-hostname-for-clients=localhost
{noformat}

Notice the use of {{-Dcom.sun.management.jmxremote.rmi.port=1099}}. This forces 
all "ephemeral" RMI client stub ports to be the specified port. With this 
config I can connect to the locator from the docker host. {{gfsh -e "connect 
locator=localhost[10334]"}}.

However, this limits the use of gfsh to the docker host. When I try and replace 
{{localhost}}, in the above gfsh command, with my host's local IP, I get 
similar behavior to the original bug. The problem appears to be that there is 
still some use of ephemeral ports which prevents this. Of interest is debugging 
into {{MBeanProcessController.getJMXServiceURL}} and the eventual {{connect}} 
deep inside RMI. More investigation is required...

> gfsh start locator command prints endless dots when starting inside docker
> --------------------------------------------------------------------------
>
>                 Key: GEODE-1935
>                 URL: https://issues.apache.org/jira/browse/GEODE-1935
>             Project: Geode
>          Issue Type: Bug
>            Reporter: Abhijat Upadhyay
>         Attachments: docker-gemfire8.zip
>
>
> I have built a docker container using gemfire 8.2.1.2. When I try to start 
> locator it keeps printing endless dots. I read about this @ 
> https://discuss.pivotal.io/hc/en-us/articles/223525728-GFSH-hang-issue-when-using-gfsh-start-or-gfsh-status-command
>  and thus using gemfire 8.2.1.2. I raised a ticket for Pivotal as well 
> https://support.pivotal.io/tickets/38301 but since it is an issue I see 
> inside docker, and they do not officially support dockerized gemfire yet, I 
> thought I will report it here as well.
> I see that you have containerized gemfire as well on github source repo but 
> not sure if you have seen this issue. I can provide you with my source code 
> as well if that helps in reproducing this issue.
> Any help/guidance is much appreciated.
> Thank you,
> AU
> ======================
> I have attached my docker project with this ticket. And instructions to run 
> it are given in the "Comments" section below.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to