pt-table-checksum --recursion-method=cluster tries to use the galera port to connect to

Bug #1220579 reported by Kenny Gryp
18
This bug affects 3 people
Affects Status Importance Assigned to Milestone
Percona Toolkit moved to https://jira.percona.com/projects/PT
Invalid
High
Frank Cizmich

Bug Description

[root@svrentpsql01 bin]# ./pt-table-checksum --recursion-method cluster -u percona -ptempaccess
Cannot connect to P=4567,h=172.29.126.138,p=...,u=percona, discovered through SHOW STATUS LIKE 'wsrep\_incoming\_addresses': DBI connect(';host=172.29.126.138;port=4567;mysql_read_default_group=client','percona',...) failed: Wrong or unknown protocol at ./pt-table-checksum line 1586

Cannot connect to P=4567,h=172.29.126.137,p=...,u=percona, discovered through SHOW STATUS LIKE 'wsrep\_incoming\_addresses': DBI connect(';host=172.29.126.137;port=4567;mysql_read_default_group=client','percona',...) failed: Wrong or unknown protocol at ./pt-table-checksum line 1586

Cannot connect to P=4567,h=172.29.126.140,p=...,u=percona, discovered through SHOW STATUS LIKE 'wsrep\_incoming\_addresses': DBI connect(';host=172.29.126.140;port=4567;mysql_read_default_group=client','percona',...) failed: Wrong or unknown protocol at ./pt-table-checksum line 1586

Use of uninitialized value in string ne at ./pt-table-checksum line 9206.

Why does mysql want to connect to port 4567? This is a galera connection port and should not be used to connect to mysql.

Revision history for this message
Kenny Gryp (gryp) wrote :

The tool then also quits as reported in bug 1210537

tags: added: i35005
tags: added: crash percona-35005 pt-table-checksum pxc
removed: i35005
Changed in percona-toolkit:
assignee: nobody → Daniel Nichter (daniel-nichter)
importance: Undecided → High
status: New → Confirmed
Changed in percona-toolkit:
milestone: none → 2.2.5
tags: added: slave-recursion
Revision history for this message
Daniel Nichter (daniel-nichter) wrote :

"Use of uninitialized value in string ne at ./pt-table-checksum line 9206." was fixed in bug 1210537.

Changed in percona-toolkit:
status: Confirmed → In Progress
Revision history for this message
Daniel Nichter (daniel-nichter) wrote :

Kenny,

What does SHOW STATUS LIKE 'wsrep\_incoming\_addresses' return?

Revision history for this message
Daniel Nichter (daniel-nichter) wrote :

And do you have PTDEBUG output? And what version of PXC?

tags: removed: crash
Revision history for this message
Kenny Gryp (gryp) wrote :

PXC version was: '5.5.30-log Percona XtraDB Cluster (GPL), wsrep_23.7.4.r3843'

Revision history for this message
Jeroen Moors (jeroen-moors) wrote :

mysql> SHOW STATUS LIKE 'wsrep\_incoming\_addresses';
+--------------------------+-------------------------------------------------------------+
| Variable_name | Value |
+--------------------------+-------------------------------------------------------------+
| wsrep_incoming_addresses | 172.29.126.138:4567,172.29.126.137:4567,172.29.126.140:4567 |
+--------------------------+-------------------------------------------------------------+
1 row in set (0.00 sec)

Revision history for this message
Daniel Nichter (daniel-nichter) wrote :

wsrep_incoming_address "Shows the comma-separated list of incoming node addresses in the cluster." What exactly does this mean? Of course the IP is for the remote, i.e. incoming node, but is the port also the remote's port or is it the local port? Plus, like you said, it's for Galera so nodes shouldn't be using it, so why is wsrep_incoming_address showing it here?

Changed in percona-toolkit:
milestone: 2.2.5 → none
Revision history for this message
Kenny Gryp (gryp) wrote :

This is very odd Daniel, Some other PXC installations I checked did not have :4567, but :3306 in there.
Maybe it's a PXC/Galera bug?

tags: added: i35005
removed: percona-35005
Changed in percona-toolkit:
assignee: Daniel Nichter (daniel-nichter) → Frank Cizmich (frank-cizmich)
Revision history for this message
Frank Cizmich (frank-cizmich) wrote :

Changing status to incomplete since it's been so long, haven't been able to reproduce, and we haven't had feedback concerning configuration files.

Changed in percona-toolkit:
status: In Progress → Incomplete
Revision history for this message
Nilnandan Joshi (nilnandan-joshi) wrote :

Again tried to check with pt-table-checksum 2.2.10 but getting this error.

[root@percona-pxc55-1 mysql]# pt-table-checksum --recursion-method cluster --user=root --password=root
Diffs cannot be detected because no cluster nodes were found. Please read the --recursion-method documentation for information.
percona-pxc55-1 is a cluster node but no other nodes or regular replicas were found. Use --recursion-method=dsn to specify the other nodes in the cluster.
[root@percona-pxc55-1 mysql]#

Even though two nodes are already connected.

| wsrep_local_state | 4 |
| wsrep_local_state_comment | Synced |
| wsrep_cert_index_size | 2 |
| wsrep_causal_reads | 0 |
| wsrep_incoming_addresses | 192.168.43.60:3306,192.168.43.61:3306 |
| wsrep_cluster_conf_id | 2 |
| wsrep_cluster_size | 2 |
| wsrep_cluster_state_uuid | eb3828f9-2125-11e4-b71b-3ff545ea7ce6 |
| wsrep_cluster_status | Primary |
| wsrep_connected | ON |
| wsrep_local_bf_aborts | 0 |
| wsrep_local_index | 0 |
| wsrep_provider_name | Galera |
| wsrep_provider_vendor | Codership Oy <email address hidden> |
| wsrep_provider_version | 2.11(r318911d) |
| wsrep_ready | ON |
| wsrep_thread_count | 2 |

Revision history for this message
Nilnandan Joshi (nilnandan-joshi) wrote :

Checked with PXC 5.5.39. Created another bug for above thing.
https://bugs.launchpad.net/percona-toolkit/+bug/1367156

Revision history for this message
Valerii Kravchuk (valerii-kravchuk) wrote :

wsrep-related setting are like these:

wsrep_sst_receive_address = zzz.yyy.xxx.155
wsrep_node_address = zzz.yyy.xxx.155:4567
wsrep_node_incoming_address = zzz.yyy.xxx.155:4567
wsrep_cluster_address = gcomm://zzz.yyy.xxx.155,zzz.yyy.xxx.156,zzz.yyy.xxx.157
wsrep_cluster_name = renta_prd_cluster_new
wsrep_provider = /usr/lib64/libgalera_smm.so
wsrep_slave_threads = 8
wsrep_sst_method = xtrabackup-v2
wsrep_sst_auth = "root:<PASSWORDBLANKED>"
wsrep_causal_reads = 1

Revision history for this message
Przemek (pmalkowski) wrote :

Couple of related bug reports were fixed since then, like:
https://bugs.launchpad.net/percona-toolkit/+bug/1217466
https://bugs.launchpad.net/percona-toolkit/+bug/1399789

I also double checked with latest toolkit and all works as expected to me:

[root@percona3 ~]# pt-table-checksum --recursion-method=cluster --databases=test --tables a,b
Not checking replica lag on percona2 because it is a cluster node.
Not checking replica lag on percona1 because it is a cluster node.
            TS ERRORS DIFFS ROWS CHUNKS SKIPPED TIME TABLE
10-31T12:28:04 0 0 3 1 0 0.072 test.a
10-31T12:28:04 0 0 3 1 0 0.057 test.b
[root@percona3 ~]# pt-table-checksum --version
pt-table-checksum 3.0.4

[root@percona3 ~]# mysql -E -e "select @@version,@@version_comment"
*************************** 1. row ***************************
        @@version: 5.7.18-15-57
@@version_comment: Percona XtraDB Cluster (GPL), Release rel15, Revision 7693d6e, WSREP version 29.20, wsrep_29.20

Tested also against PXC 5.6.36

Changed in percona-toolkit:
status: Incomplete → Invalid
Revision history for this message
Przemek (pmalkowski) wrote :

@Valerii, btw, in the variables you mentioned in example, this one is incorrect:
wsrep_node_incoming_address = zzz.yyy.xxx.155:4567
as this variable defines address and port for client, so MySQL connections, hence it should be 3306 (or better just empty).
For reference:
http://galeracluster.com/documentation-webpages/mysqlwsrepoptions.html#wsrep-node-incoming-address

Revision history for this message
Shahriyar Rzayev (rzayev-sehriyar) wrote :

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PT-368

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.