Isilon: As mensagens /var/log em alguns nós mostram mensagens de memória insuficiente de tempos em tempos

Summary: As mensagens /var/log em alguns nós mostram essas mensagens de falta de memória (OOM) de tempos em tempos: OOM: v_wire_count: 2832982, v_active_count: 516

This article applies to This article does not apply to This article is not tied to any specific product. Not all product versions are identified in this article.

Symptoms

 Um conjunto completo de mensagens tem a seguinte aparência: 
2021-08-17T14:40:50.703082+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: OOM: v_wire_count: 2843879, v_active_count: 871
2021-08-17T14:40:50.703246+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: Malloc Pigs:
2021-08-17T14:40:50.703277+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: Type                   InUse   MemUse   Requests
2021-08-17T14:40:50.703303+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: 8kB dinodes           840507  647775K 13330759531
2021-08-17T14:40:50.703325+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: isi_hash               31424   88002K 1548035744
2021-08-17T14:40:50.703344+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: lbm super              17535   84150K    1435677
2021-08-17T14:40:50.703362+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: layout_hints          183473   57336K   22320273
2021-08-17T14:40:50.703379+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: iaddr_set             886694   55419K 5188138062
2021-08-17T14:40:50.703397+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: devbuf                 17379   37840K     184744
2021-08-17T14:40:50.703414+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: newblk                    26   32774K    2364548
2021-08-17T14:40:50.703432+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: bar_owner_vec259         261   32288K    2220148
2021-08-17T14:40:50.703451+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: parent_vec153         218013   27538K    4567646
2021-08-17T14:40:50.703475+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: ddvec84               183471   22840K  177809736
2021-08-17T14:40:50.703496+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: inodedep                  22   16390K    1677902
2021-08-17T14:40:50.703515+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: sysctloid             201426   10989K     201598
2021-08-17T14:40:50.703535+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: vfscache                   4    8241K          4
2021-08-17T14:40:50.703555+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: KTR                        2    7168K          2
2021-08-17T14:40:50.703574+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: ptr_llcb_map           56126    7017K  116151979
2021-08-17T14:40:50.703592+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: Unshown bins account for 84039K
2021-08-17T14:40:50.703611+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: Total: 1219801K
2021-08-17T14:40:50.703630+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: UMA Zalloc Pigs:
2021-08-17T14:40:50.703648+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: ZONE NAME       SIZE     LIMIT     COUNT   MEM USED
2021-08-17T14:40:50.703668+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: mbuf          000256, 06012268, 00017191, 00004297K
2021-08-17T14:40:50.703687+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: socket        000872, 00765000, 00002327, 00001981K
2021-08-17T14:40:50.703704+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: tcpcb         001080, 00765000, 00001652, 00001742K
2021-08-17T14:40:50.703722+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: tcp_inpcb     000488, 00765000, 00001665, 00000793K
2021-08-17T14:40:50.703740+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: udp_inpcb     000488, 00765000, 00000270, 00000128K
2021-08-17T14:40:50.703758+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: unpcb         000192, 00765000, 00000391, 00000073K
2021-08-17T14:40:50.703776+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: hostcache     000096, 00015360, 00000191, 00000017K
2021-08-17T14:40:50.703794+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: udpcb         000032, 00765000, 00000270, 00000008K
2021-08-17T14:40:50.703812+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: clpbuf        001016, 00015872, 00000006, 00000005K
2021-08-17T14:40:50.703832+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: tcptw         000096, 00027767, 00000013, 00000001K
2021-08-17T14:40:50.703853+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: ripcb         000488, 00765000, 00000001, 00000000K
2021-08-17T14:40:50.703872+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: Unshown zones account for 0K
2021-08-17T14:40:50.703890+12:00 <0.4> cluster-1(id1) /boot/kernel.amd64/kernel: Total: 9051K

Cause

Uma nova função do OneFS introduzida no OneFS 9.2 adiciona essas mensagens. 
Essa nova função registra informações de uso de memória sempre que uma função de pageout é chamada pelo pagedaemon quando não consegue colocar a memória livre acima de free_target.
O sistema operacional está chamando proativamente a função acima para registrar informações caso a memória fique ainda mais baixa e o nó pare de responder ou entre em pane.

Resolution

Importante:
Se o nó tiver apenas 16 GB de RAM, oriente o usuário a fazer upgrade para 64 GB.

Se o nó tiver 24 GB de RAM ou mais, essas mensagens poderão ser ignoradas com segurança se todas estas condições forem atendidas:

  • O /var/log/vmlog mostra que a memória livre não ficava frequentemente abaixo do destino de 50% livre
  • Nenhum nó do cluster entra em reinicialização de pane com BUF_TIMELOCK ou outras mensagens de tempo de espera excedido relacionadas à OOM na string de pane
  • A pasta freevnodes sysctl Mostra habilitado (valor de 1) em todos os nós:
# isi_for_array 'sysctl vfs.vnlru_reuse_freevnodes'

Nota:
Lembre-se de alterar sysctl vfs.vnlru_reuse_freevnodes de volta ao valor padrão (0 - zero) se todas as seguintes condições forem atendidas:
  • O nó tem 24 GB ou mais de memória física
  • A pasta vnlru_reuse_freevnodes=1 foi originalmente adicionada para resolver o problema de Dinodes de 8 KB
  • A versão do OneFS é atualizada para um destes níveis ou posterior:
    • OneFS 9.2.1.25_GA-RUP_2023-12
    • OneFS 9.4.0.17_GA-RUP_2024-02
    • OneFS 9.5.0.7_LTS2023_GA-RUP (janeiro de 2024)
Article Properties
Article Number: 000191515
Article Type: Solution
Last Modified: 03 Apr 2024
Version:  5
Find answers to your questions from other Dell users
Support Services
Check if your device is covered by Support Services.