#57 Failover testing

[Tathagat Swain]

Hi,

I am testing the failover of datanodes in case of PostgresXL.

I have 2 servers :
Server 1 has : 1GTM , 1 Coordinator and 1 Datanode
Server 2 has : 1 Datanode.

When I shut down data node 2 , I am not able to insert data using the coordinator.
it shows "ERROR: Failed to get pooled connections"

Is this normal or I have made some mistakes in my configuration.?
has anyone else faced this problem .?

[Koichi Suzuki]

To failover datanodes, you need slave for each of them. XL itself is
not for failover system. You can configure slaves using log-shipping
replication of PostgreSQL. Pgxc_ctl will help to configure them as
well and provides command interface for failover. Please note that
pgxc_ctl does not provide automatic failover or failure monitoring.

Regards;

Koichi Suzuki

2015-04-10 22:37 GMT+09:00 Tathagat Swain tathagatswain@users.sf.net:

---

[tickets:#57] Failover testing

Status: open
Milestone: 9.2rc
Created: Fri Apr 10, 2015 01:37 PM UTC by Tathagat Swain
Last Updated: Fri Apr 10, 2015 01:37 PM UTC
Owner: nobody

Hi,

I am testing the failover of datanodes in case of PostgresXL.

I have 2 servers :
Server 1 has : 1GTM , 1 Coordinator and 1 Datanode
Server 2 has : 1 Datanode.

When I shut down data node 2 , I am not able to insert data using the
coordinator.
it shows "ERROR: Failed to get pooled connections"

Is this normal or I have made some mistakes in my configuration.?
has anyone else faced this problem .?

---

Sent from sourceforge.net because you indicated interest in
https://sourceforge.net/p/postgres-xl/tickets/57/

To unsubscribe from further messages, please visit
https://sourceforge.net/auth/subscriptions/

[Tathagat Swain]

Thanks Koichi.

One more thing , so if a node goes down for any reason and there are no slave for data-nodes , the whole cluster goes down , until either you bring up the crashed node or remove it altogether from the cluster ?

[Koichi Suzuki]

Good question. So far, there could be some problem if you have a
distributed table. You may encounter errors due to this failed node
when you try to remove it from the cluster. If all the tables are
replicated, you may be able to remove the failed node safely by "ALTER
TABLE" and "DROP NODE".

I've not tested this case yet. Appreciate if you test this.

Regards;

Koichi Suzuki

2015-04-13 13:05 GMT+09:00 Tathagat Swain tathagatswain@users.sf.net:

Thanks Koichi.

One more thing , so if a node goes down for any reason and there are no
slave for data-nodes , the whole cluster goes down , until either you bring
up the crashed node or remove it altogether from the cluster ?

---

[tickets:#57] Failover testing

Status: open
Milestone: 9.2rc
Created: Fri Apr 10, 2015 01:37 PM UTC by Tathagat Swain
Last Updated: Fri Apr 10, 2015 01:37 PM UTC
Owner: nobody

Hi,

I am testing the failover of datanodes in case of PostgresXL.

I have 2 servers :
Server 1 has : 1GTM , 1 Coordinator and 1 Datanode
Server 2 has : 1 Datanode.

When I shut down data node 2 , I am not able to insert data using the
coordinator.
it shows "ERROR: Failed to get pooled connections"

Is this normal or I have made some mistakes in my configuration.?
has anyone else faced this problem .?

---

Sent from sourceforge.net because you indicated interest in
https://sourceforge.net/p/postgres-xl/tickets/57/

To unsubscribe from further messages, please visit
https://sourceforge.net/auth/subscriptions/