I'm trying to find out if and how Hazelcast manages version-updates of nodes in a cluster. I found this:
Support Hazelcast update in running cluster
It seems to be possible just with minor version updates...
Has someone experienced updating a Hazelcast Cluster? What should I do if I want to update a major version change?
I wasn't able to find something about updating cluster nodes in the Hazelcast documentation...
Thanks for any help,
Christian
Currently the only way is to totally shut down the cluster, replace the jars and restart. Rolling upgrade is on the agenda.
Related
We are looking to upgrade our RDS Mysql From 5.7 to 8.0.11. When try to modify and upgrade the instance from AWS UI we get a
Database instance is in a state that cannot be upgraded: PreUpgrade checks failed: RDS detected Incompatibilities upgrading to MySQL 8.0.15. More details can be found in the PrePatchCompatibility.log file, accessible in the 'logs' section below.
error in the recent events area but when we download the PrePatchCompatibility.log the only messages we see are
3) Usage of utf8mb3 charset
The following objects use the utf8mb3 character set. It is recommended to convert them to use utf8mb4 instead, for improved Unicode support.
More Information:
https://dev.mysql.com/doc/refman/8.0/en/charset-unicode-utf8mb3.html
schema.table.column - column's default character set: utf8
I've looked this message up and the compatibility issues associated with it and this is supposed to be a warning. The bottom of the log shows
Errors: 0
Database Objects Affected: 261
From everything I've Read this should work.
Has anyone successfully upgraded from mysql 5.7 to 8.0 in aws RDS? Would I actually have to change all of my columns from utf8mb3 to utf8mb4 or is there something else going on?
After contacting AWS support team, they disabled pre-upgrade checks for our instance. After which we were able to upgrade successfully from 5.7 to 8.0.
They have informed that AWS has a bug and patch for that is ready and will be released soon. Though they didn't provide any ETA for the same. But in case you are in hurry, you can contact them ask for the private patch (or disabling pre-upgrade checks in this case).
RDS MySQL Engineer here.
This post was sent to us by an RDS customer, so I thought it would be interesting to chime in and provide you some visibility about this issue. Our team is aware of this issue, and a patch deployment was scheduled to address it. This fix will be transparent to all of our RDS MySQL customers, and will require no additional action from your side once it is released.
I would like apologize for any inconvenience this may have caused you and reiterate that we appreciate you bringing this issue to our attention, we have listened :)
With that said this may be a good time to remind you that our team is constantly monitoring our AWS Official channels such as #aws (Twitter), AWS RDS MySQL Forums (forums.aws.amazon.com/forum.jspa?forumID=60) and our support tickets of course.
We recommend you using these channels if you are looking for a faster response from us.
Best regards, Your friends from the RDS MySQL team.
Hi I am trying to create a wordpress application using docker compose and I use the Galera MariaDB catalog entry from the rancher.
I can get all the set up working fine. I use external links and connect to the load balancer with some environment variable like this:
external_links:
r-galera_galera-lb_1:mysql
I can see the tables being replicated in the cluster, however if I reboot the machine, even after the stack becomes active again, I fail to launch the application.
I get the error like this:
> wordpress-docker-php-fpm | MySQL "CREATE DATABASE" Error: WSREP has not yet prepared node for application use
> wordpress-docker-php-fpm exited with code 1
When I remove the whole Galera Stack and make a new one I get my wordpress setup working again.
I had to come to this forum for this issue since I couldn't contact any maintainer of the catalog (there isn't any contact info). Can someone help in this regard ?
Hello Syed Alam Abbas,
the issue with your approach is = the cluster is not properly shut down and started. If you reboot your machine, the clusters nodes will go out of sync and store the latest state they are in. If you restart the machine and everything is back online - you have an unsynced cluster. You can follow this guide to recover your cluster.
The guide is pretty straight forward:
check the latest state of all your nodes "wsrep_last_committed" with SHOW STATUS LIKE 'wsrep_%';
Promote the node which has the most-up-to-date data to be the primary.
I am looking at creating a high availability Wordpress installation across two servers with a load balancer to distribute the traffic between these servers, I intend to setup a LAMP stack on these two web servers and configure MySQL master master replication to ensure both servers have the same content similar to the setup at https://www.linode.com/docs/websites/cms/high-availability-wordpress
For anyone created a similar setup before, is there anything to watch out for? Will this work without there being differences in the database assuming both nodes are online? If one node goes down will this synchronise with the other master node successfully?
This is a pretty old post, but I've set up exactly this recently.
I do not recommend setting up WordPress with master-master replication. The nodes went out of sync multiple times and resulted in split-brain. We've reverted back to master-slave replication because of the instability of the setup.
I recently tried this and regret it. I realize your post is now almost 2 years old, but thought I should put in my 2 cents for anyone considering this approach. I recommend you go with MariaDB with MaxScale to separate the reads and writes, and go with master-slave configuration. I'm running into all sorts of deadlocks, and issues with this configuration. Currently using MariaDB with Galera, with HAProxy up front.
Please can anyone help me to disable "innodb_doublewrite" for my MySQL database hosted at Amazon RDS. I need this as we need to quickly update around 15 Million rows.
I know there is a startup option for this:
--skip-innodb_doublewrite
But how to use it?
Apart from that, Amazon RDS Parameters Group does not show the option of "innodb_doublewrite" for editing. Amazon does not also allow direct editing of my.cnf file.
I can access the MySQL through my Linux Server. But don't know exactly how to use the startup option for Amazon RDS. Please can any one help me to disable this option?
AWS support confirmed the following as of August 2, 2015:
"innodb_doublewrite can't be modified in RDS MySQL instances"
RDS is a managed database service that provides some abstractions to managing a database directly on EC2 but in exchange you lose some flexibility including what options you are able to apply to your database instance.
All start up parameters that would normally be specified in my.cnf should instead be applied via the RDS Parameter Groups but, as you mention in your question, innodb_doublewrite is not one of the configurable parameters.
If the data you are updating is not likely to be updated during normal operation you could always attempt to make the update on a read slave (SET GLOBAL READ_ONLY=0 first) and then promote the read replica and point your application at this instance.
I'm currently using tomcat7 configured with it's own jdbc-pool as described here: http://www.tomcatexpert.com/blog/2010/04/01/configuring-jdbc-pool-high-concurrency .
I also have a master/master configuration for mysql (replicated) and I was wondering on how to (if possible..) configure tomcat to failover between the 2 mysql's . so for example if mysql1 goes down, open all the pool again against mysql2 .
Any ideas on how to achieve that?
I also don't mind switching to another jdbc implementation if not tomcat's.
You can try C-JDBC and its doc.
HA-JDBC seems to fit the bill. Its a library which provides high availability and fail-over features among many listed on the website. Connection pooling is one aspect which needs to be looked upon.