Frequently Asked Questions¶
- How do you solve locking issues like auto-increment?
- What if one of the nodes crashes and InnoDB recovery rolls back some transactions?
- How can I check the Galera node health?
- How does Percona XtraDB Cluster handle big transactions?
- Is it possible to have different table structures on the nodes?
- What if a node fails or there is a network issue between nodes?
- How would the quorum mechanism handle split brain?
- I have a two nodes setup. When node1 fails, node2 does not accept commands, why?
- Is it possible to set up a cluster without state transfer?
- What TCP ports are used by Percona XtraDB Cluster?
- Is there “async” mode or only “sync” commits are supported?
- Does it work with regular MySQL replication?
- Why doesn’t the init script (/etc/init.d/mysql) start?
- What does “nc: invalid option – ‘d’” in the sst.err log file mean?
How do you solve locking issues like auto-increment?¶
For auto-increment, Percona XtraDB Cluster changes
auto_increment_offset for each new node. In a single-node workload, locking is handled in the same way as InnoDB. In case of write load on several nodes, Percona XtraDB Cluster uses optimistic locking and the application may receive lock error in the response to the
What if one of the nodes crashes and InnoDB recovery rolls back some transactions?¶
When a node crashes, after restart it will copy the whole dataset from another node (if there were changes to data since the crash).
How can I check the Galera node health?¶
To check the health of a Galera node, use the following query:
SELECT 1 FROM dual;
The following results of the previous query are possible:
- You get the row with
id=1(node is healthy)
- Unknown error (node is online but Galera is not connected/synced with the cluster)
- Connection error (node is not online)
You can also check a node’s health with the
clustercheck script. First set up the
GRANT USAGE ON *.* TO 'clustercheck'@'localhost' IDENTIFIED BY PASSWORD '*2470C0C06DEE42FD1618BB99005ADCA2EC9D1E19';
You can then check a node’s health by running the
/usr/bin/clustercheck clustercheck password 0
If the node is running correctly you should get the following status:
HTTP/1.1 200 OK Content-Type: text/plain Connection: close Content-Length: 40 Percona XtraDB Cluster Node is synced.
In case node isn’t sync or if it is offline, status will look like:
HTTP/1.1 503 Service Unavailable Content-Type: text/plain Connection: close Content-Length: 44 Percona XtraDB Cluster Node is not synced.
clustercheck script has the following syntax:
<user> <pass> <available_when_donor=0|1> <log_file> <available_when_readonly=0|1> <defaults_extra_file>
server_args = user pass 1 /var/log/log-file 0 /etc/my.cnf.local
server_args = user pass 1 /var/log/log-file 1 /etc/my.cnf.local
How does Percona XtraDB Cluster handle big transactions?¶
Percona XtraDB Cluster populates write set in memory before replication, and this sets the limit for the size of transactions that make sense. There are wsrep variables for maximum row count and maximum size of write set to make sure that the server does not run out of memory.
Is it possible to have different table structures on the nodes?¶
For example, if there are four nodes, with four tables:
sessions_d. If you want each table in a separate node, this is not possible for InnoDB tables. However, it will work for MEMORY tables.
What if a node fails or there is a network issue between nodes?¶
The quorum mechanism in Percona XtraDB Cluster will decide what nodes can accept traffic and will shut down the nodes that do not belong to the quorum. Later when the failure is fixed, the nodes will need to copy data from the working cluster.
The algorithm for quorum is Dynamic Linear Voting (DLV). The quorum is preserved if (and only if) the sum weight of the nodes in a new component strictly exceeds half that of the preceding Primary Component, minus the nodes which left gracefully.
The mechanism is described in detail in Galera documentation.
How would the quorum mechanism handle split brain?¶
It would not. If there is no way to decide on the primary component, Percona XtraDB Cluster has no way to resolve a split brain. The minimal recommendation is to have 3 nodes. However, it is possibile to allow a node to handle the traffic with the following option:
wsrep_provider_options="pc.ignore_sb = yes"
I have a two nodes setup. When node1 fails, node2 does not accept commands, why?¶
Is it possible to set up a cluster without state transfer?¶
It is possible in two ways:
- By default, Galera reads starting position from a text file
<datadir>/grastate.dat. Simply make this file identical on all nodes, and there will be no state transfer upon start.
- Use the
wsrep_start_positionvariable to start the nodes with the same
What TCP ports are used by Percona XtraDB Cluster?¶
You may need to open up to four ports if you are using a firewall:
Regular MySQL port (default is 3306).
Port for group communication (default is 4567). It can be changed using the following option:
wsrep_provider_options ="gmcast.listen_addr=tcp://0.0.0.0:4010; "
Port for State Snaphot Transfer (default is 4444). It can be changed using the following option:
Port for Incremental State Transfer (default is port for group communication + 1 or 4568). It can be changed using the following option:
wsrep_provider_options = "ist.recv_addr=10.11.12.206:7777; "
Is there “async” mode or only “sync” commits are supported?¶
Percona XtraDB Cluster does not support “async” mode, all commits are synchronous on all nodes. To be precise, the commits are “virtually” synchronous, which means that the transaction should pass “certification” on nodes, not physical commit. “Certification” means a guarantee that the transaction does not have conflicts with other transactions on the corresponding node.
Does it work with regular MySQL replication?¶
Yes. On the node you are going to use as master, you should enable
Why doesn’t the init script (/etc/init.d/mysql) start?¶
Try to disable SELinux with the following command:
$ echo 0 > /selinux/enforce