ESXi6.5上的Ubuntu虛機在遠程SSH時宕機

狀況是這樣的, 有一個ESXi6.5上跑的Ubuntu虛機, 版本是18.04.1, 今天升級成18.04.2後, 就發現遠程鏈接SSH不對勁, 在本地登陸和操做都好好的, 只要遠程SSH一鏈接, 服務器立馬奔潰. 開始覺得是SSHD配置問題, 修改各類參數沒效果, 而系統/var/log/syslog裏面也沒有任何相關的錯誤信息. 後來實在沒辦法, 重裝了, 仍是不行. 懷疑是這個虛機的文件壞了, 把虛機刪了, 再新建虛機重裝, 發現仍是這個問題.

可是在重建虛機安裝後, 鏈接SSH時在本地的命令行看到了一個頗有價值的錯誤輸出

[84978.843130] kernel BUG at drivers/net/vmxnet3/vmxnet3_drv.c:1441!
[84978.843167] invalid opcode: 0000 [#1] SMP
[84978.843191] Modules linked in: vmw_vsock_vmci_transport(E) vsock(E) xt_conntrack(E) iptable_mangle(E) iptable_nat(E) nf_conntrack_ipv4(E) nf_defrag_ipv4(E) nf_nat_ipv4(E) nf_nat(E) iptable_filter(E) ip_tables(E) xt_LOG(E) nf_conntrack(E) coretemp(E) hwmon(E) kvm_intel(E) kvm(E) irqbypass(E) mousedev(E) hid_generic(E) aesni_intel(E) vmw_balloon(E) aes_x86_64(E) glue_helper(E) lrw(E) gf128mul(E) ablk_helper(E) evdev(E) cryptd(E) psmouse(E) usbhid(E) hid(E) nfit(E) intel_agp(E) vmw_vmci(E) battery(E) i2c_piix4(E) intel_gtt(E) acpi_cpufreq(E) tpm_tis(E) tpm_tis_core(E) tpm(E) ac(E) button(E) sch_fq_codel(E) crc32c_intel(E) uhci_hcd(E) ehci_pci(E) ehci_hcd(E) usbcore(E) usb_common(E) autofs4(E)

在VMWare的官網上查到了對應的bug和解決方案 https://kb.vmware.com/s/article/2151480

This issue occurs due to a bug in vmxnet3 vNIC backend which is part of the vmkernel. This issue occurs if all the below are true:

1. Linux VM is running kernel >= 4.8
2. HW version of VM is >=13
3. ESXi version is 6.5

This is a known issue affecting VMware ESXi 6.5.
This issue is resolved in VMware ESXi 6.5 Update 1, available at VMware Downloads.
To work around this issue if you do not want to upgrade, use any one of these options.
1. Add the vmxnet3.rev.30 = FALSE parameter in the vmx file of virtual machine:
2. Power off the virtual machine.
3. Edit the vmx file and add the below parameter:
    vmxnet3.rev.30 = FALSE
4. Power on the virtual machine.

If you do not want to power off the virtual machine, disable the receive data ring for each vmxnet3 vNIC on the VM by running this command:
    ethtool -G ethX rx-mini 0
Note: Replace ethX with virtual machine interface name.