Symptoms

• Database dump fails with the following error:

  # plesk db dump example_db > /var/www/vhosts/example.com/example.sql
  signal: killed
  mysqldump: Got errno 32 on write

• The following entries are found in /var/log/messages:

  test kernel: systemd invoked oom-killer: gfp_mask=0x201da, order=0, oom_score_adj=0
  test kernel: systemd cpuset=/ mems_allowed=0

  test kernel: lowmem_reserve[]: 0 0 0 0
  test kernel: Node 0 DMA: 3_4kB (M) 6_8kB (UEM) 7_16kB (UEM) 2_32kB (UM) 2_64kB (M) 1_128kB (E) 2_256kB (EM) 1_512kB (E) 2_1024kB (UE) 2_2048kB (EM) 0_4096kB = 7660kB
  test kernel: Node 0 DMA32: 283_4kB (UEM) 734_8kB (UE) 550_16kB (UE) 297_32kB (UEM) 222_64kB (UEM) 22_128kB (EM) 1_256kB (M) 0_512kB 0_1024kB 1_2048kB (M) 0_4096kB = 44636kB
  test kernel: Node 0 hugepages_total=0 hugepages_free=0 hugepages_surp=0 hugepages_size=2048kB
  test kernel: 58470 total pagecache pages
  test kernel: 0 pages in swap cache
  test kernel: Swap cache stats: add 0, delete 0, find 0/0
  test kernel: Free swap = 0kB
  test kernel: Total swap = 0kB
  test kernel: 524153 pages RAM

• Dumping the database with the native MySQL command is successful:

  # MYSQL_PWD=cat /etc/psa/.psa.shadow mysqldump -uadmin example_db > /var/www/vhosts/example.com/example.sql

   

• Memory on the server might be somewhat limited, e.g:

  # free -mh
  total used free shared buff/cache available
  Mem: 1.8G 734M 157M 235M 946M 696M
  Swap: 0B 0B 0B

Cause

The 'plesk db' command uses more memory than a native MySQL dump command. As a result, the process is being killed by OOM killer.

Resolution

Apply one or more of the following solutions:

1. Adjust OOM settings

2. Increase RAM on the server

3. Use native mysqldump:

   # MYSQL_PWD=cat /etc/psa/.psa.shadow mysqldump -uadmin example_db > /var/www/vhosts/example.com/example.sql