Skip to content

Conversation

@dbaba
Copy link

@dbaba dbaba commented Jul 12, 2015

I'd found an issue when creating a ZK ensemble with 3 ZK containers (by setting MYID and SERVERS) along with dnsdock.
The issue is the first of the three containers kept failing to resolve other hosts (UnknownHostException was thrown) which started after the first one is running. This was caused by Java's DNS cache preventing the ZK from performing DNS queries, so I disabled it.

After disabling the DNS cache, ZK ensemble on docker worked fine.

Please consider to review and accept this request if the change is helpful for you as well or discard it otherwise.

@gavinwhyte
Copy link
Member

Hello @dbaba thank you for reporting this issue, I will investigate this issue and get back to you, as I have not experienced this problem previously.

@bric3
Copy link

bric3 commented Feb 4, 2016

Playing with JVM network cache should not help in this situation much as zookeeper have his own way to deal with this, and there's an issue on that matter that exists from quite some time (https://issues.apache.org/jira/browse/ZOOKEEPER-1506). Should have been released with 3.4.7, but it had a critical bug (a deadlock), so it should be available in 3.4.8.

@stephanlindauer
Copy link

is this resolved? we are still having issues with that, using gcr.io/google_samples/k8szk:v2

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants