I'm planning to upgrade MariaDB Galera cluster from 10.1 to 10.2. Does anyone have details for steps to upgrade? My idea is something
Backup
Shutdown cluster
Uninstall 10.1 from each node
Install 10.2 to each node
Run mysql_upgrade at node which going to started first
Configure the first node and start
Configure rest of nodes and start them
I have three node cluster with maxscale loadbalancing.
You can upgrade the cluster in a rolling fashion, i.e. one node at a time without shutting down the others. That is one of the benefits of Galera cluster.
Make sure to avoid 10.2.9 or be ready to edit mysqld_safe, see here.
For each node:
maxadmin: set server $node-name maintenance
Backup databases and config files
Shutdown the mysqld instance
Uninstall 10.1. On Redhat use rpm -e --nodeps rather than yum remove to avoid uninstalling packages such as postfix and cronie.
Install 10.2
Copy back config files, change any mariadb-10.1 sections to mariadb-10.2
Startup the mysqld instance
If you're on Redhat, CentOS or Fedora run mysql_upgrade
maxadmin: clear server $node-name maintenance
Related
I'm a student working on a test cluster, consisting of around 25 hosts. We installed using Ambari and have FreeIpa running on a host as a dns and ldap server. The rest are typical Hadoop
infrastructure. Hive was failing and I wondered whether the db connection parameters used during the Ambari installation were incorrect and I tried to find a way to re-run the db connection process. I didn't get anywhere and it was late so I left it, ambari interface working.
Next morning, ambari webUI seems to be down. I thought that maybe the webserver needed restarted so I tried the following:
[akidd#dw ~]$ sudo ambari-server start
Using python /usr/bin/python
Starting ambari-server
ERROR: Exiting with exit code 1.
REASON: Unable to detect a system user for Ambari Server.
- If this is a new setup, then run the "ambari-server setup" command to create the user
- If this is an upgrade of an existing setup, run the "ambari-server upgrade" command.
Refer to the Ambari documentation for more information on setup and upgrade.
Can anyone help me to understand what could have happened?
If I run ambari-server setup will the existing cluster be ok assuming I create everything like for like with how it was originally?
Thanks for your help!
#user3535074 You should try to start it with the user that installed it.
If you do run ambari-server setup as current user, remember to choose No the following options:
Customize user account for ambari-server daemon [y/n] (n)? n
Do you want to change Oracle JDK [y/n] (n)? n
Enter advanced database configuration [y/n] (n)? n
More info on the following post, including how to backup ambari database before running setup again:
https://community.cloudera.com/t5/Support-Questions/Ambari-server-failed-to-start-after-system-reboot-Below-is/td-p/203806
When I try to start Cassandra with brew services start cassandra, I get a message that it has been successfully started. However, when I do brew services list, the Cassandra status is started but the started text is orange (not green).
It seems that Cassandra is not running at all. I am unable to connect to it with cqlsh and get no results when I run ps aux | grep cass.
Any suggestions how to troubleshoot Cassandra not starting from Homebrew?
This is a problem with Java version - existing releases of the Cassandra do work only with Java 8. Support for Java 9,10,11, ... is added only in Cassandra 4.0 that isn't released yet. You need to install Java 8 and point Cassandra to it.
You can read more about the changes in following blog post.
Hi,
We are trying to install CDH cluster on Redhat 7 remote server using cloudera-installer.bin file, in standalone mode( we have only 1 host) . We are specifying hostname/ip address of the machine during installation , it is able to resolve it. But the installation halts during parcel distribution stage. Here are the logs of cloudera-scm-agent :(We tried both cloudera express edition and entrerprise trial version too)
['http://INHUSZ1-V250152:7180/cmf/parcel/download/CDH-5.15.1-1.cdh5.15.1.p0.4-el7.parcel'] location=/opt/cloudera/parcels/.flood/CDH-5.15.1-1.cdh5.15.1.p0.4-el7.parcel progress=0]
[03/Oct/2018 10:11:55 +0000] 28315 Thread-13 downloader INFO Current state: CDH-5.15.1-1.cdh5.15.1.p0.4-el7.parcel [totalDownloaded=0 totalSize=2120090032 upload=0 state=downloading seed=['http://INHUSZ1-V250152:7180/cmf/parcel/download/CDH-5.15.1-1.cdh5.15.1.p0.4-el7.parcel'] location=/opt/cloudera/parcels/.flood/CDH-5.15.1-1.cdh5.15.1.p0.4-el7.parcel progress=0]
[03/Oct/2018 10:11:57 +0000] 28315 Thread-13 downloader INFO Current state: CDH-5.15.1-1.cdh5.15.1.p0.4-el7.parcel [totalDownloaded=0 totalSize=2120090032 upload=0 state=downloading seed=['http://INHUSZ1-V250152:7180/cmf/parcel/download/CDH-5.15.1-1.cdh5.15.1.p0.4-el7.parcel'] location=/opt/cloudera/parcels/.flood/CDH-5.15.1-1.cdh5.15.1.p0.4-el7.parcel progress=0]
[03/Oct/2018 10:11:59 +0000] 28315 Thread-13 downloader INFO Current state: CDH-5.15.1-1.cdh5.15.1.p0.4-el7.parcel [totalDownloaded=0 totalSize=2120090032 upload=0 state=downloading seed=['http://INHUSZ1-V250152:7180/cmf/parcel/download/CDH-5.15.1-1.cdh5.15.1.p0.4-el7.parcel'] location=/opt/cloudera/parcels/.flood/CDH-5.15.1-1.cdh5.15.1.p0.4-el7.parcel progress=0]
Please let us know what can be done
I just had the same error message and stall during install at parcel distribution stage.
Installing a single node (test) cluster on CentOS 7.5 with CDH Express 5.15.
Solution that worked for me was adding the node IP and FQDN to /etc/hosts (previously it only contained entries for 127.0.0.1 localhost):
[root#mynode ~]# vi /etc/hosts
192.168.1.1 myhostname.mydomain
Then restarted Cloudera SCM Agent:
[root#mynode ~]# service cloudera-scm-agent restart
Installation then continued successfully.
Do the following:
Stop all services.
Deactivate all in-use parcels.
Shut down the Cloudera Manager Agent on all hosts.
Move the existing parcels to the new location.
Configure the host parcel directory.
Start the Cloudera Manager Agents.
Activate the parcels.
Start all services.
Delete the corresponding parcels package from below folder including .torrent file
/opt/cloudera/parcels/.flood/
Download and distribute
This is happening because .torrent file is corrupted
ERROR [main] 2017-08-04 13:24:21,949 CassandraDaemon.java:638 - Detected unreadable sstables /opt/cassandra/data/some_key_space/ep_lc_events-adc44160dbe611e6953689bcd3ed73aa/mc-547-big-Summary.db, and many others...
That has happened after I upgraded Cassandra to 3 version and after a while downgraded it to 2nd version.
When I run this command: sudo service cassandra status
I have got such message:
could not access pidfile for Cassandra
In /var/log/cassandra/system.log I have logs which I wrote at the beginning.
PS: let me pay your attention that everything is happening on EC2 Amazon instance.
Well, I have just upgraded back to 3rd version, used cassandra-unloader to export all data, then downgraded back to 2nd version and used cassandra-loader to import all data. But if you were lucky and had backups and snapshots it would not be an obstacle for you.
PS. Afterwards, I had to run this command nodetool resetlocalschema to reset local schema and resynchronize.
PPS. This you can find how to do that.
https://github.com/brianmhess/cassandra-loader
I also got the same error, but it was due to switching between cassandra 4.0.0 and version 3.11 and back again while using docker.
Update the version to the right one for the ssltable, or delete the data volume:
docker-compose logs cassandra
docker volumes ls
docker ps
docker-compose down
docker volume rm testapp_cassandra
docker volume ls
docker-compose up
I am now trying to install ES on windows 2016 Nano server, it works when start ES in command line, but failed to start ES as windows service, maybe due to some changes on Nano server, but I have no hint at all why it didn't work.
Anyone has experience on installing ES on Nano server?
or Where can I get the source code of ES windows service, elasticsearch-windows-x64.exe? It could be helpful to troubleshoot the root cause.
I haven't got a chance to test it on Nano Server 2016, but I've installed ES 5.1.2 + JDK1.8.0_121 on Windows 2016 Server Core successfully and was able to start the service.
Install JDK and set SYSTEM ENV with setx -m JAVA_HOME
"C:\PROGRA~1\Java\jdk1.8.0_121" command. It doesn't work with user level ENV variable such as using set JAVA_HOME=xxx.
unzip ES to c:\elk\elasticsearch
Edit the elasticsearch.yml config file, config the Cluster and Node names and save it
install the ES service C:\elk\elasticsearch\bin>elasticsearch-service.bat install
start the ES service C:\elk\elasticsearch\bin>elasticsearch-service.bat start
Reference:
https://github.com/elastic/elasticsearch/issues/18373
http://blog.technotesdesk.com/setting-a-system-environment-variable-from-command-line-in-windows-server-2012-r2/