garbd requires port number in sysconfig

[Vadym Chepkov]

Comments in /etc/sysconfig/garb indicates that adding port is optional:

# A comma-separated list of node addresses (address[:port]) in the cluster

And yet, if you won't add the port, garb fails to start:

GALERA_NODES="10.121.215.31,10.121.215.32"

Oct 6 11:59:57 wpq001 systemd: Starting Galera Arbitrator Daemon...
Oct 6 11:59:57 wpq001 systemd: Started Galera Arbitrator Daemon.
Oct 6 11:59:57 wpq001 garb-systemd: Starting garbd
Oct 6 11:59:57 wpq001 garb-systemd: 2015-10-06 11:59:57.100 INFO: CRC-32C: using hardware acceleration.
Oct 6 11:59:57 wpq001 garb-systemd: 2015-10-06 11:59:57.100 INFO: Read config:
Oct 6 11:59:57 wpq001 garb-systemd: daemon: 0
Oct 6 11:59:57 wpq001 garb-systemd: name: garb
Oct 6 11:59:57 wpq001 garb-systemd: address: gcomm://10.121.215.31,10.121.215.32
Oct 6 11:59:57 wpq001 garb-systemd: group: wpq-cluster
Oct 6 11:59:57 wpq001 garb-systemd: sst: trivial
Oct 6 11:59:57 wpq001 garb-systemd: donor:
Oct 6 11:59:57 wpq001 garb-systemd: options: gcs.fc_limit=9999999; gcs.fc_factor=1.0; gcs.fc_master_slave=yes
Oct 6 11:59:57 wpq001 garb-systemd: cfg:
Oct 6 11:59:57 wpq001 garb-systemd: log:
Oct 6 11:59:57 wpq001 garb-systemd: 2015-10-06 11:59:57.101 INFO: protonet asio version 0
Oct 6 11:59:57 wpq001 garb-systemd: 2015-10-06 11:59:57.101 INFO: Using CRC-32C for message checksums.
Oct 6 11:59:57 wpq001 garb-systemd: 2015-10-06 11:59:57.101 INFO: backend: asio
Oct 6 11:59:57 wpq001 garb-systemd: 2015-10-06 11:59:57.102 WARN: access file(./gvwstate.dat) failed(No such file or directory)
Oct 6 11:59:57 wpq001 garb-systemd: 2015-10-06 11:59:57.102 INFO: restore pc from disk failed
Oct 6 11:59:57 wpq001 garb-systemd: 2015-10-06 11:59:57.102 INFO: GMCast version 0
Oct 6 11:59:57 wpq001 garb-systemd: 2015-10-06 11:59:57.102 FATAL: Exception in creating receive loop.
Oct 6 11:59:57 wpq001 systemd: garb.service: main process exited, code=exited, status=1/FAILURE
Oct 6 11:59:57 wpq001 systemd: Unit garb.service entered failed state.

Percona-XtraDB-Cluster-garbd-3-3.12-1.rhel7.x86_64

If you add port, service starts normally:

GALERA_NODES="10.121.215.31:4567,10.121.215.32:4567"

[Ramesh Sivaraman (rameshvs02)]

Verified it and able to reproduce the issue.

2015-10-12 02:20:32.662 INFO: CRC-32C: using "slicing-by-8" algorithm.
2015-10-12 02:20:32.662 INFO: Read config:
        daemon: 1
        name: garb
        address: gcomm://10.0.0.243
        group: pxc-pquery
        sst: trivial
        donor:
        options: gcs.fc_limit=9999999; gcs.fc_factor=1.0; gcs.fc_master_slave=yes
        cfg:
        log: /tmp/garbd.log

2015-10-12 02:20:32.665 INFO: protonet asio version 0
2015-10-12 02:20:32.665 INFO: Using CRC-32C for message checksums.
2015-10-12 02:20:32.665 INFO: backend: asio
2015-10-12 02:20:32.667 INFO: restore pc from disk successfully
2015-10-12 02:20:32.671 INFO: GMCast version 0
2015-10-12 02:20:32.672 FATAL: Exception in creating receive loop.

[Antonio (toniusco)]

Same problem in debian,

but in debian not work anymore with ports and without ports.

[Antonio (toniusco)]

Last version that work in debian: percona-xtradb-cluster-garbd-3.x 3.8.3390.wheezy

[Krunal Bauskar (krunal-bauskar)]

commit 5a196d055c51b2a01260ec0330016159f3282b2a
Merge: af09c7a ee98aa2
Author: Krunal Bauskar <email address hidden>
Date: Wed Dec 9 08:47:30 2015 +0530
Merge pull request #38 from natsys/3.x-pxc-484

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PXC-1858