Clusterkommunikation gestört?

[fpausp]

Hab bei einem Node immer wieder Aussetzer:


pve01:


pve02:


pvestor01:


Kann mir bitte jemand einen Tipp geben wie ich das eingrenzen kann?

 

[fpausp]

OK hab etwas verdächtiges am 1. Node pve01 gefunden:

May  8 12:01:47 pve01 kernel: [ 1275.025544] vmbr0: port 1(enp0s25) entered blocking state
May  8 12:01:47 pve01 kernel: [ 1275.025547] vmbr0: port 1(enp0s25) entered forwarding state
May  8 12:02:43 pve01 kernel: [ 1331.256813] vmbr0: port 1(enp0s25) entered disabled state
May  8 12:02:49 pve01 kernel: [ 1337.061809] e1000e: enp0s25 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
May  8 12:02:49 pve01 kernel: [ 1337.061851] vmbr0: port 1(enp0s25) entered blocking state
May  8 12:02:49 pve01 kernel: [ 1337.061854] vmbr0: port 1(enp0s25) entered forwarding state
May  8 12:03:37 pve01 kernel: [ 1385.269059] vmbr0: port 1(enp0s25) entered disabled state
May  8 12:03:43 pve01 kernel: [ 1390.974134] e1000e: enp0s25 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
May  8 12:03:43 pve01 kernel: [ 1390.974175] vmbr0: port 1(enp0s25) entered blocking state
May  8 12:03:43 pve01 kernel: [ 1390.974177] vmbr0: port 1(enp0s25) entered forwarding state
May  8 12:06:02 pve01 kernel: [ 1529.905194] nfs: server 192.168.xxx.xxx not responding, timed out
May  8 12:07:15 pve01 kernel: [ 1603.382326] vmbr0: port 1(enp0s25) entered disabled state
May  8 12:07:21 pve01 kernel: [ 1609.051316] e1000e: enp0s25 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
May  8 12:07:21 pve01 kernel: [ 1609.051357] vmbr0: port 1(enp0s25) entered blocking state
May  8 12:07:21 pve01 kernel: [ 1609.051360] vmbr0: port 1(enp0s25) entered forwarding state
May  8 12:09:02 pve01 kernel: [ 1709.938169] nfs: server 192.168.xxx.xxx not responding, timed out
May  8 12:09:02 pve01 kernel: [ 1710.390987] vmbr0: port 1(enp0s25) entered disabled state
May  8 12:09:07 pve01 kernel: [ 1715.058215] nfs: server 192.168.xxx.xxx not responding, timed out
May  8 12:09:08 pve01 kernel: [ 1716.111900] e1000e: enp0s25 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
May  8 12:09:08 pve01 kernel: [ 1716.111939] vmbr0: port 1(enp0s25) entered blocking state
May  8 12:09:08 pve01 kernel: [ 1716.111942] vmbr0: port 1(enp0s25) entered forwarding state
May  8 12:10:14 pve01 kernel: [ 1782.331547] vmbr0: port 1(enp0s25) entered disabled state
May  8 12:10:20 pve01 kernel: [ 1788.096317] e1000e: enp0s25 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
May  8 12:10:20 pve01 kernel: [ 1788.096357] vmbr0: port 1(enp0s25) entered blocking state
May  8 12:10:20 pve01 kernel: [ 1788.096360] vmbr0: port 1(enp0s25) entered forwarding state
May  8 12:11:15 pve01 kernel: [ 1843.255642] vmbr0: port 1(enp0s25) entered disabled state
May  8 12:11:21 pve01 kernel: [ 1848.932635] e1000e: enp0s25 NIC Link is Up 1000 Mbps Full Duplex, Flow Control: None
May  8 12:11:21 pve01 kernel: [ 1848.932674] vmbr0: port 1(enp0s25) entered blocking state
May  8 12:11:21 pve01 kernel: [ 1848.932677] vmbr0: port 1(enp0s25) entered forwarding state

 

[fpausp]

Konnte das Problem wie folgt entschärfen:

ethtool -K enp0s25 gso off gro off tso off

network-nic-offloads

Plus:

GRUB_CMDLINE_LINUX_DEFAULT="quiet pcie_aspm=off"

in der /boot/grub/grub.cfg