Shulou(Shulou.com)06/02 Report--

Environment zabbix 4 + pgpool + postgres 11.0 cluster + Centos 7.5 + Python2.7.5 [3.6]

The cause of the failure is PG Master disk space 150G. Due to the large number of zabbix monitoring devices (about 1000 points and more template monitoring items), the archived files in the database grow very fast, usually eating up a G space in about five minutes. After the disk space is consumed, the service of pg&pgpool will be stopped, which will affect the use of zabbix system. To solve this problem, bash shell scripts and crontab tasks are created to clean up archive files on a regular basis.

Fault log screenshot:

PostgreSQL error PANIC: no valid checkpoint logging solution was found: 1. Log in to postgres: docker run-ti postgres / bin/bash [not a docker environment to skip this step directly] 2. Switch to postgres user 3. Execute the repair command: / usr/pgsql-11/bin/pg_resetwal-f / var/lib/pgsql/11/data [depending on the environment and installation, the file path may be different, you can use the locate or find command to search for pg_resetwal] [versions of pg_resetxlog-f DATADIR postgres less than 10.0 can use this command;-f enforce updates] 4. If you see "Write-ahead log reset", the repair is successful. # systemctl start postgresql-11.service # systemctl status postgresql-11.service # systemctl start pgpool-II-11.service # systemctl status pgpool-II-11.service # netstat-pltn

At this time, the service of PG is up, but the database may not work properly. Continue to use the command to check.

Su-postgres-bash-4.2$ psql psql: FATAL: xlog flush request 399/FCA1D7D8 is not satisfied-flushed only to 399/E720DE18 CONTEXT: writing block 2225 of relation base/16385/17835-- this error in this Times can be patiently waited for a period of time (about ten minutes). It takes some time for both pg and pgpool to recover from failure before they can be used normally. -bash-4.2$ psql psql Type "help" for help. It is recommended to use pgpooladmin tool to repair postgres cluster with postgres=#.

Note: just click "return / disconnect" & "reload" indicated by the red arrow, and the cluster business can be resumed, which is easier to maintain than command line management.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.