[MDEV-9472] 13 node Galera cluster blocks all writes until affected node restart - Jira

XML

Word

Printable

Details

Type: Bug
Status: Closed (View Workflow)
Priority: Major
Resolution: Incomplete
Affects Version/s: 10.1.10
Fix Version/s: N/A
Component/s: Galera, Locking
Labels:
- galera
Environment:

Hide
13 nodes running MariaDB 10.1.10 with Galera
This occurs with the shipping .9 version of the Galera library as well as the latest .14 version.

Layout:

DC1: 3 nodes
DC2: 3 nodes
DC3: 3 nodes
DC4: 3 nodes
DC5: 1 node

Each site minus DC5 has at least 4 async slaves hanging off a VIP, where only one of the 3 nodes at that site is active. (Priority LB method)

Show
13 nodes running MariaDB 10.1.10 with Galera This occurs with the shipping .9 version of the Galera library as well as the latest .14 version. Layout: DC1: 3 nodes DC2: 3 nodes DC3: 3 nodes DC4: 3 nodes DC5: 1 node Each site minus DC5 has at least 4 async slaves hanging off a VIP, where only one of the 3 nodes at that site is active. (Priority LB method)

Description

My 13 node cluster is used for DNS and therefor has a lot more reads than writes. The issue seems to stem from DNS slave nodes in each site which perform a lot of UPDATE statements on the same table. Each site has a VIP where only one of the 3 nodes at that site are active at one time. I will attach my error log and configuration file.

Attachments

- Sort By Name
- Sort By Date
- Ascending
- Descending
- Thumbnails
- List
- Download All

v-mysql-003-net.ams2.symcpe.net.err.bz2
6.43 MB
2016-01-26 21:53
v-mysql-003-net.lhr1.symcpe.net.err.bz2
9.62 MB
2016-01-26 21:53

Activity

People

Assignee:: Jan Lindström (Inactive)

Reporter:: Bryan Seitz

Votes:: 0 Vote for this issue

Watchers:: 3 Start watching this issue

Dates

Created:: 2016-01-26 21:27

Updated:: 2019-12-12 12:21

Resolved:: 2019-12-12 12:21

Git Integration

Error rendering 'com.xiplink.jira.git.jira_git_plugin:git-issue-webpanel'. Please contact your Jira administrators.