Problem with importing/deploying MariaDB Galera from WebUI
Env:
Controller: 1.7.6.3940, Frontend: 1.7.6.6868-#e56aa7 (Ubuntu 18.04.4)
DB: Ubuntu 18.04.4, MariaDB 10.4
Problem:
After successful importing of 3nodes MariaDB Galera cluster I get critical error: No contact to the management node / MGMD_NO_CONTACT. Moreover, sometimes ClusterControl shows my cluster's vendor as Oracle. As a result, my cluster is always shown in critical state, however CC gets all system information from nodes. Yet another thing: on Manage/Hosts view all my database nodes are always in unknown state.
Also, in /var/log/cmon_ndbworker.log:
2020-05-22T18:01:47.676-02:00 : (INFO) NDB connection: 127.0.0.1:1186
2020-05-22T18:01:47.676-02:00 : (INFO) cmon_ndbworker 1.7.6.3940 starting up.
2020-05-22T18:01:47.676-02:00 : (INFO) 127.0.0.1:1186: Connect error: Failed to connect MGM (code: 1011): Could not connect to socket.
In /var/log/cmon_1.log:
2020-05-22T18:01:53.921-02:00 : (INFO) infosec-db01:3306: Vendor auto-detection succeed from version strings: mariadb/10.4.
2020-05-22T18:01:53.922-02:00 : (INFO) infosec-db02:3306: Vendor auto-detection succeed from version strings: mariadb/10.4.
2020-05-22T18:01:53.922-02:00 : (INFO) infosec-db03:3306: Vendor auto-detection succeed from version strings: mariadb/10.4.
2020-05-22T18:01:53.956-02:00 : (ERROR) Table 'ndbinfo.memoryusage' doesn't exist: SELECT node_id, memory_type, total_pages, used_pages, round(total) as total_bytes, round(used) as used_bytes, total, used FROM ndbinfo.memoryusage GROUP BY node_id, memory_type ORDER BY node_id, memory_type
2020-05-22T18:01:53.956-02:00 : (ERROR) Table 'ndbinfo.logbuffers' doesn't exist: SELECT node_id, log_type, log_id, log_part, sum(total), sum(used) FROM ndbinfo.logbuffers GROUP BY log_type, node_id
2020-05-22T18:01:53.957-02:00 : (ERROR) Table 'ndbinfo.logspaces' doesn't exist: SELECT node_id, log_type, log_id, log_part, sum(total), sum(used) FROM ndbinfo.logspaces GROUP BY log_type, node_id
2020-05-22T18:01:53.958-02:00 : (ERROR) Table 'ndbinfo.counters' doesn't exist: SELECT node_id, lcase(counter_name), sum(val) FROM ndbinfo.counters WHERE counter_name not like '<unknown>' GROUP BY counter_name, node_id ORDER BY node_id, counter_name
2020-05-22T18:01:53.959-02:00 : (ERROR) Table 'ndbinfo.nodes' doesn't exist: SELECT node_id, uptime, start_phase FROM ndbinfo.nodes
2020-05-22T18:01:53.960-02:00 : (ERROR) Table 'ndbinfo.diskpagebuffer' doesn't exist: SELECT node_id, block_instance, pages_written, pages_written_lcp, pages_read, log_waits, page_requests_direct_return, page_requests_wait_queue, page_requests_wait_io, IFNULL(round(100 * page_requests_direct_return / (page_requests_direct_return + page_requests_wait_io + page_requests_wait_queue)),0) FROM ndbinfo.diskpagebuffer
Possible cause:
It seems like CC faulty decides that I have MySQL Cluster (NDB) with management node, not Galera Cluster. But I'm sure that I chose right options during importing nodes. Also I tried to manually set vendor to "mariadb" in cluster settings, but with no results. I've also tried new deployment with three clean ubuntu hosts. Same result.
-
FYI,
I've successfully imported Galera cluster with cli:
s9s cluster --register \
--cluster-type=galera \
--nodes="node01;node02;node03" \
--vendor=mariadb \
--provider-version=10.4 \
--db-admin="root" \
--db-admin-passwd="password" \
--os-user=osuser \
--cluster-name="SomeCluster" \
--waitNo idea why I can't do this from Web UI. I tried both Chrome and Firefox browsers with no luck. It looks like Web UI did not take my parameters while importing cluster with wizard and create a new one with partially both MySQL NDB and Galera options.
Please sign in to leave a comment.
Comments
3 comments