Recovering From MySQL Cluster Downtime
Page last updated:
This topic describes how to bootstrap your MySQL cluster in the event of a cluster failure.
You can bootstrap your cluster by using one of two methods:
- Run the bootstrap errand. See Run the Bootstrap Errand below.
- Bootstrap manually. See Bootstrap Manually below.
When to Bootstrap
You must bootstrap a cluster that loses quorum. A cluster loses quorum when fewer than half the nodes can communicate with each other for longer than the configured grace period. If a cluster does not lose quorum, individual unhealthy nodes automatically rejoin the cluster after resolving the error, restarting the node, or restoring connectivity.
To check whether your cluster has lost quorum, look for the following symptoms:
All nodes appear “Unhealthy” on the proxy dashboard. To log in to the proxy dashboard, see Log In to Proxy Dashboard below.
All responsive nodes report the value of
wsrep_cluster_status
asnon-Primary
:mysql> SHOW STATUS LIKE 'wsrep_cluster_status'; +----------------------+-------------+ | Variable_name | Value | +----------------------+-------------+ | wsrep_cluster_status | non-Primary | +----------------------+-------------+
All unresponsive nodes respond with
ERROR 1047
when using most statement types in the MySQL client:mysql> select * from mysql.user; ERROR 1047 (08S01) at line 1: WSREP has not yet prepared node for application use
Log In to Proxy Dashboard
To log in to the proxy dashboard:
Get the proxy dashboard credentials by entering the following URL into your web browser and recording the output:
https://OPS-MAN-FQDN/api/v0/deployed/products/TAS-FOR-VMS-GUID/variables?name=mysql-proxy-dashboard-credentials
Where:
OPS-MAN-FQDN
is the fully-qualified domain name (FQDN) for your Ops Manager deployment.TAS-FOR-VMS-GUID
is the GUID of your TAS for VMs tile. For more information about finding your GUID, see the Ops Manager API documentation
Log in to the proxy dashboard by entering the following URL into your web browser:
https://BOSH-JOB-INDEX-proxy-p-mysql-ert.YOUR-SYSTEM-DOMAIN/
Where:
BOSH-JOB-INDEX
is0
,1
, or2
, representing the three proxies deployed on each node.SYSTEM-DOMAIN
is the System domain you configured in the Domains pane of the BOSH Director tile.
Run the Bootstrap Errand
Ops Manager includes a BOSH errand that automates the manual bootstrapping procedure in the Bootstrap Manually section below.
It finds the node with the highest transaction sequence number and asks it to start up by itself (i.e. in bootstrap mode) and then asks the remaining nodes to join the cluster.
In most cases, running the errand recovers your cluster. But, certain scenarios require additional steps.
Determine Type of Cluster Failure
To determine which set of instructions to follow:
List your MySQL instances by running:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT instances
Where:
YOUR-ENV
is the environment where you deployed the cluster.YOUR-DEPLOYMENT
is the deployment cluster name.
For example:
$ bosh -e prod -d mysql instances
Find and record the
Process State
for your MySQL instances. In the following example output, the MySQL instances are in thefailing
process state.Instance Process State AZ IPs backup-restore/c635410e-917d-46aa-b054-86d222b6d1c0 running us-central1-b 10.0.4.47 bootstrap/a31af4ff-e1df-4ff1-a781-abc3c6320ed4 - us-central1-b - broker-registrar/1a93e53d-af7c-4308-85d4-3b2b80d504e4 - us-central1-b 10.0.4.58 cf-mysql-broker/137d52b8-a1b0-41f3-847f-c44f51f87728 running us-central1-c 10.0.4.57 cf-mysql-broker/28b463b1-cc12-42bf-b34b-82ca7c417c41 running us-central1-b 10.0.4.56 deregister-and-purge-instances/4cb93432-4d90-4f1d-8152-d0c238fa5aab - us-central1-b - monitoring/f7117dcb-1c22-495e-a99e-cf2add90dea9 running us-central1-b 10.0.4.48 mysql/220fe72a-9026-4e2e-9fe3-1f5c0b6bf09b failing us-central1-b 10.0.4.44 mysql/28a210ac-cb98-4ab4-9672-9f4c661c57b8 failing us-central1-f 10.0.4.46 mysql/c1639373-26a2-44ce-85db-c9fe5a42964b failing us-central1-c 10.0.4.45 proxy/87c5683d-12f5-426c-b925-62521529f64a running us-central1-b 10.0.4.60 proxy/b0115ccd-7973-42d3-b6de-edb5ae53c63e running us-central1-c 10.0.4.61 rejoin-unsafe/8ce9370a-e86b-4638-bf76-e103f858413f - us-central1-b - smoke-tests/e026aaef-efd9-4644-8d14-0811cb1ba733 - us-central1-b 10.0.4.59
Choose your scenario:
- If your MySQL instances are in the
failing
state, continue to Scenario 1. - If your MySQL instances are in the
-
state, continue to Scenario 2. - If your MySQL instances are in the
stopped
state, continue to Scenario 3.
- If your MySQL instances are in the
Scenario 1: VMs Running, Cluster Disrupted
In this scenario, the VMs are running, but the cluster has been disrupted.
To bootstrap in this scenario:
Run the bootstrap errand on the VM where the bootstrap errand is co-located, by running:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT run-errand --instance=VM bootstrap
Where:
YOUR-ENV
is the name of your environment.YOUR-DEPLOYMENT
is the name of your deployment.VM
isclock_global/0
if you use TAS for VMs andcontrol/0
if you use Small Footprint TAS for VMs.
Note: The errand runs for a long time, during which no output is returned.
The command returns many lines of output, eventually followed by:
Bootstrap errand completed [stderr]
- echo 'Started bootstrap errand ...'
- JOB_DIR=/var/vcap/jobs/bootstrap
- CONFIG_PATH=/var/vcap/jobs/bootstrap/config/config.yml
- /var/vcap/packages/bootstrap/bin/cf-mysql-bootstrap -configPath=/var/vcap/jobs/bootstrap/config/config.yml
- echo 'Bootstrap errand completed'
- exit 0 Errand 'bootstrap' completed successfully (exit code 0)
If the errand fails, run the bootstrap errand command again after a few minutes. The bootstrap errand might not work the first time.
If the errand fails after several tries, bootstrap your cluster manually. See Bootstrap Manually below.
Scenario 2: VMs Terminated or Lost
In severe circumstances, such as a power failure, it is possible to lose all your VMs. You must re-create them before you can begin recovering the cluster.
When MySQL instances are in the -
state, the VMs are lost. The procedures in
this scenario bring the instances from a -
state to a failing
state.
Then you run the bootstrap errand similar to Scenario 1
above and restore configuration.
To recover terminated or lost VMs, follow the procedures in the sections below:
Re-create the Missing VMs: Bring MySQL instances from a
-
state to afailing
state.Run the Bootstrap Errand: Because your instances are now in the
failing
state, continue similarly to Scenario 1 above.Restore the BOSH Configuration: Go back to unignoring all instances and redeploy. This is a critical and mandatory step.
Caution:
If you do not set each of your ignored instances to unignore
,
your instances are not updated in future deploys.
You must follow the procedure in the final section of Scenario 2,
Restore the BOSH Configuration.
Re-create the Missing VMs
The procedure in this section uses BOSH to re-create the VMs, install software on them, and try to start the jobs.
The procedure below allows you to:
Redeploy your cluster while expecting the jobs to fail.
Instruct BOSH to ignore the state of each instance in your cluster. This allows BOSH to deploy the software to each instance even if the instance is failing.
To re-create your missing VMs:
If BOSH Resurrector is activated, deactivate it by running:
bosh -e YOUR-ENV update-resurrection off
Where
YOUR-ENV
is the name of your environment.Download the current manifest by running:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT manifest > /tmp/manifest.yml
Where:
YOUR-ENV
is the name of your environment.YOUR-DEPLOYMENT
is the name of your deployment.
Redeploy deployment by running:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT deploy /tmp/manifest.yml
Where:
YOUR-ENV
is the name of your environment.YOUR-DEPLOYMENT
is the name of your deployment.
Note: Expect one of the MySQL VMs to fail. Deploying causes BOSH to create new VMs and install the software. Forming a cluster is in a subsequent step.
View the instance GUID of the VM that attempted to start by running:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT instances
Where:
YOUR-ENV
is the name of your environment.YOUR-DEPLOYMENT
is the name of your deployment.
Record the instance GUID, which is the string after
mysql/
in your BOSH instances output.Instruct BOSH to ignore each MySQL VM by running:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT ignore mysql/INSTANCE-GUID
Where:
YOUR-ENV
is the name of your environment.YOUR-DEPLOYMENT
is the name of your deployment.INSTANCE-GUID
is the GUID of your instance you recorded in the above step.
Repeat steps 4 and 5 until all instances have attempted to start.
If you deactivated the BOSH Resurrector in step 2, re-enable it by running:
bosh -e YOUR-ENV update-resurrection on
Where
YOUR-ENV
is the name of your environment.Confirm that your MySQL instances have gone from the
-
state to thefailing
state by running:bosh -e YOUR-ENV -d YOUR-DEPLOYMENT instances
Where:
YOUR-ENV
is the name of your environment.YOUR-DEPLOYMENT
is the name of your deployment.
Run the Bootstrap Errand
After you re-create the VMs, all instances now have a failing
process state and have the MySQL code.
You must run the bootstrap errand to recover the cluster.
To bootstrap:
Run the bootstrap errand by running:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT run-errand --instance=VM bootstrap
Where:
YOUR-ENV
is the name of your environment.YOUR-DEPLOYMENT
is the name of your deployment.VM
isclock_global/0
if you use TAS for VMs andcontrol/0
if you use Small Footprint TAS for VMs.
Note: The errand runs for a long time, during which no output is returned.
The command returns many lines of output, eventually with the following successful output:
Bootstrap errand completed [stderr] echo 'Started bootstrap errand ...' JOB_DIR=/var/vcap/jobs/bootstrap CONFIG_PATH=/var/vcap/jobs/bootstrap/config/config.yml /var/vcap/packages/bootstrap/bin/cf-mysql-bootstrap -configPath=/var/vcap/jobs/bootstrap/config/config.yml echo 'Bootstrap errand completed' exit 0 Errand 'bootstrap' completed successfully (exit code 0)
If the errand fails, run the bootstrap errand command again after a few minutes. The bootstrap errand might not work immediately.
See that the errand completes successfully in the shell output and continue to Restore the BOSH Configuration below.
Note: After you complete the bootstrap errand, you might still see instances in the
failing
state. Continue to the next section anyway.
Restore the BOSH Configuration
Caution: If you do not set each of your ignored instances
to unignore
, your instances are never updated in future deploys.
To restore your BOSH configuration to its previous state, this procedure unignores each instance that was previously ignored:
For each ignored instance, run:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT unignore mysql/INSTANCE-GUID
Where:
YOUR-ENV
is the name of your environment.YOUR-DEPLOYMENT
is the name of your deployment.INSTANCE-GUID
is the GUID of your instance.
Redeploy your deployment by running:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT deploy
Where:
YOUR-ENV
is the name of your environment.YOUR-DEPLOYMENT
is the name of your deployment.
Verify that all
mysql
instances are in arunning
state by running:bosh -e YOUR-ENV -d YOUR-DEPLOYMENT instances
Where:
YOUR-ENV
is the name of your environment.YOUR-DEPLOYMENT
is the name of your deployment.
Bootstrap Manually
If the bootstrap errand is not able to automatically recover the cluster, you need to do the steps manually.
Follow the procedures in the sections below to manually bootstrap your cluster.
Caution: The following procedures are prone to user-error and can result in lost data if followed incorrectly. Follow the procedure in Bootstrap with the BOSH Errand above first, and only resort to the manual process if the errand fails to repair the cluster.
Shut Down MySQL
Follow these steps to stop the galera-init
process for each node in the cluster.
For each node, record if the monit stop
command was successful:
SSH into the node using the procedure in Advanced Troubleshooting with the BOSH CLI.
To shut down the
mysqld
process on the node, run:monit stop galera-init
Record if the monit command succeeds or exits with an error:
- If monit succeeds in stopping
galera-init
, then you can use monit to restart this node. Follow all the steps below including the steps marked Monit Restart but omitting the steps marked Manual Redeploy. If monit exits with the following error, then you must manually deploy this node:
Warning: include files not found '/var/vcap/monit/job/*.monitrc' monit: action failed -- There is no service by that name
Follow all the steps below including the steps marked Manual Redeploy but omitting the steps marked Monit Restart.
- If monit succeeds in stopping
Repeat the steps above for each node in the cluster.
You cannot bootstrap the cluster unless you have shut down the
mysqld
process on all nodes in the cluster.
Determine which Node to Bootstrap
To avoid losing data, you must bootstrap from a node in the cluster that has
the highest transaction sequence number (seqno
).
For each node in the cluster, to find its seqno
:
SSH into the node using the procedure in Advanced Troubleshooting with the BOSH CLI.
Find
seqno
values in the node’s Galera state file,grastate.dat
by running:cat /var/vcap/store/pxc-mysql/grastate.dat | grep 'seqno:'
Do one of the following, based on the last and highest
seqno
value in thegrep
output above:- If
seqno
is a positive integer, then the node shut down gracefully. Record this number for comparison with the latestseqno
of other nodes in the cluster. If
seqno
is-1
, then the node crashed or was stopped. Proceed as follows to attempt to recover theseqno
from the database:Temporarily start the database and append the last sequence number to its error log by running:
/var/vcap/packages/pxc/bin/mysqld --defaults-file=/var/vcap/jobs/pxc-mysql/config/my.cnf --wsrep-recover
Scan the end of the error log
mysql.err.log
for the recoveredseqno
. It appears after the group ID (a UUID) and a colon. For example, in the following output the recoveredseqno
is15026
:$ grep "Recovered position" /var/vcap/sys/log/pxc-mysql/mysql.err.log | tail -1 150225 18:09:42 mysqld_safe WSREP: Recovered position e93955c7-b797-11e4-9faa-9a6f0b73eb46:15026
If the node never connected to the cluster before crashing, it might not have been assigned a group ID. In this case, there is nothing to recover. Do not choose this node for bootstrapping unless all the other nodes also crashed.
- If
After you retrieve the
seqno
for all nodes in your cluster, identify the one with the highestseqno
. If multiple nodes share the same highestseqno
, and it is not-1
, you can bootstrap from any of them.
Bootstrap the First Node
After determining the node with the highest seqno
, do the following
to bootstrap the node:
Note: Only run these bootstrap commands on the
node with the highest seqno
.
Otherwise the node with the highest seqno
is unable
to join the new cluster unless its data is abandoned.
Its mysqld
process exits with an error.
SSH into the node using the procedure in Advanced Troubleshooting with BOSH CLI.
Update the node state to trigger its initialization of the cluster by running:
echo -n "NEEDS_BOOTSTRAP" > /var/vcap/store/pxc-mysql/state.txt
Monit Restart: If in Shut Down MySQL above you successfully used
monit
to shut down yourgalera-init
process, then re-launch themysqld
process on the new bootstrap node.Start the
mysqld
process by running:monit start galera-init
It can take up to ten minutes for
monit
to start themysqld
process. To confirm if themysqld
process has started successfully, run:watch monit summary
If monit succeeds in starting the galera-init process, then the output includes the line
Process 'galera-init' running
.
Manual Redeploy: If in Shut Down MySQL above you encountered
monit
errors, then redeploy themysqld
software to your bootstrap node as follows:- Leave the MySQL SSH login shell and return to your local environment.
Target BOSH on your bootstrap node by instructing it to ignore the other nodes in your cluster. For nodes all nodes except the bootstrap node you identified above, run:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT ignore mysql/M bosh -e YOUR-ENV -d YOUR-DEPLOYMENT ignore mysql/N
Where
N
andM
are the numbers of the non-bootstrapped nodes. For example, if you bootstrap node 0, thenM
=1 andN
=2.Turn off the BOSH Resurrector by running:
bosh update-resurrection off
Use the BOSH manifest to bootstrap your bootstrap machine by running:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT manifest > /tmp/manifest.yml bosh -e YOUR-ENV -d YOUR-DEPLOYMENT deploy /tmp/manifest.yml
Restart Remaining Nodes
After the bootstrapped node is running, restart the remaining nodes.
The procedure that you follow for restarting a node, depends on the output you got for that node during Shut Down MySQL above. Do one of the following procedures:
Monit Restart
If in Shut Down MySQL above you successfully used monit
to shut down your
galera-init
process, then restart the nodes as follows:
SSH into the node using the procedure in Advanced Troubleshooting with BOSH CLI.
Start the
mysqld
process withmonit
by running:monit start galera-init
If the interruptor prevents the node from starting, follow the manual procedure to force the node to rejoin the cluster. See How to manually force a MySQL node to rejoin if a node cannot rejoin the HA (Galera) cluster in the Support knowledge base.
Caution: Forcing a node to rejoin the cluster is a destructive procedure. Only follow the procedure with the assistance of Support.
If the
monit start
command fails, it might be because the node with the highestseqno
ismysql/0
.In this case:
Ensure BOSH ignores updating
mysql/0
by running:bosh -e YOUR-ENV -d YOUR-DEPLOYMENT ignore mysql/0
Where:
YOUR-ENV
is the name of your environment.YOUR-DEPLOYMENT
is the name of your deployment.
Navigate to Ops Manager in a browser, log in, and click Apply Changes.
When the deploy finishes, run the following command from the Ops Manager VM:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT unignore mysql/0
Where:
YOUR-ENV
is the name of your environment.YOUR-DEPLOYMENT
is the name of your deployment.
Manual Redeploy
If in Shut Down MySQL above you encountered monit
errors,
then restart the nodes as follows:
Instruct BOSH to no longer ignore the non-bootstrap nodes in your cluster by running:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT unignore mysql/M bosh -e YOUR-ENV -d YOUR-DEPLOYMENT unignore mysql/N
Where
N
andM
are the numbers of the non-bootstrapped nodes. For example, if you bootstrap node 0, thenM
=1 andN
=2.Redeploy software to the other two nodes and have them rejoin the cluster, bootstrapped from the node above by running:
bosh -e YOUR-ENV -d YOUR-DEPLOYMENT deploy /tmp/manifest.yml
You only need to run this command once to deploy both the nodes that you unignored in the step above.
With your redeploys completed, turn the BOSH Resurrector back on by running:
bosh -e YOUR-ENV update-resurrection on
Verify that the Nodes Have Joined the Cluster
The final task is to verify that all the nodes have joined the cluster.
SSH into the bootstrap node then run the following command to output the total number of nodes in the cluster:
mysql> SHOW STATUS LIKE 'wsrep_cluster_size';