is it possible to upgrade zookeeper without having job managers restart

[Barisa Obradovic]

We are running flink 1.10 cluster, with 3 zookeepers for HA.

I noticed that trying to upgrade the zookeepers, even 1 by one, causes
jobmanagers to crash

```
java.net.UnknownHostException: foo-host: Name or service not known
        at java.net.Inet6AddressImpl.lookupAllHostAddr(Native Method)
        at java.net.InetAddress$2.lookupAllHostAddr(InetAddress.java:929)
        at java.net.InetAddress.getAddressesFromNameService(InetAddress.java:1324)
        at java.net.InetAddress.getAllByName0(InetAddress.java:1277)
        at java.net.InetAddress.getAllByName(InetAddress.java:1193)
        at java.net.InetAddress.getAllByName(InetAddress.java:1127)
```


With some google fu, we noticed https://github.com/apache/flink/pull/11938.

Is there newer version of flink we can uptake where we can upgrade
zookeepers without job manager downtime?
And is there any other way around the fact, that having even a single
zookeeper down causes the jobmanagers to crash, thus eliminating the
usefulness of HA




--
Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/

[Kenzyme]

Hello Barisa,

Seems like you asked the same question as me just ~ a week ago!

http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/ZooKeeper-connection-SUSPENDING-td38779.html

We will have to wait until the feature mentioned in the JIRA ticket is implemented.

Best,

Kenzyme Le

‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐

On Wednesday, October 21st, 2020 at 12:10 PM, Barisa Obradovic <[hidden email]> wrote:

> We are running flink 1.10 cluster, with 3 zookeepers for HA.
>
> I noticed that trying to upgrade the zookeepers, even 1 by one, causes
>
> jobmanagers to crash
>
>     java.net.UnknownHostException: foo-host: Name or service not known
>     at java.net.Inet6AddressImpl.lookupAllHostAddr(Native Method)
>     at java.net.InetAddress$2.lookupAllHostAddr(InetAddress.java:929)
>     at java.net.InetAddress.getAddressesFromNameService(InetAddress.java:1324)
>     at java.net.InetAddress.getAllByName0(InetAddress.java:1277)
>     at java.net.InetAddress.getAllByName(InetAddress.java:1193)
>     at java.net.InetAddress.getAllByName(InetAddress.java:1127)
>
>
> With some google fu, we noticed https://github.com/apache/flink/pull/11938.
>
> Is there newer version of flink we can uptake where we can upgrade
>
> zookeepers without job manager downtime?
>
> And is there any other way around the fact, that having even a single
>
> zookeeper down causes the jobmanagers to crash, thus eliminating the
>
> usefulness of HA
>
>
> ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
>
> Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/

[Barisa Obradovic]

Great, thank you :)



--
Sent from: http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/