VMware ESXi 6.0 iscsi_vmk msleep langsomt startup

En VMware ESXi 6.0 Server med build version 4192238

Problem:

En genstart af vmware-host tager helt op til 40 minutter før den er online igen.

Fejlfinding

Hvis jeg læser i /var/log/vmkernel.log, ser jeg at msleep har kørt i ca. 20 minutter:

2016-10-05T19:03:55.779Z cpu6:33410)Tcpip: 2589: msleep returned 4
2016-10-05T19:04:00.782Z cpu6:33410)Tcpip: 2589: msleep returned 4
2016-10-05T19:04:05.784Z cpu6:33410)Tcpip: 2589: msleep returned 4
2016-10-05T19:04:11.390Z cpu6:33410)Tcpip: 2589: msleep returned 4
2016-10-05T19:04:16.393Z cpu6:33410)Tcpip: 2589: msleep returned 4
2016-10-05T19:04:21.396Z cpu6:33410)Tcpip: 2589: msleep returned 4
2016-10-05T19:04:26.996Z cpu6:33410)Tcpip: 2589: msleep returned 4
....... rigtig mange gange

og jeg ser:

2016-10-05T19:05:54.747Z cpu5:32832)NMP: nmp_ThrottleLogForDevice:3298: Cmd 0x12 (0x43a5cd77d5c0, 0) to dev "naa.6a4badb00053137e000004744c03707e" on path "vmhba38:C13:T0:L5" Failed: H:0x0 D:0x2 P:0x0 Valid sense data: 0x5 0x24 0x0. Act:NONE
2016-10-05T19:05:54.808Z cpu7:33541)NMP: nmp_ThrottleLogForDevice:3298: Cmd 0x12 (0x43a5cd77d5c0, 0) to dev "naa.6842b2b00064d53a000038f64fe2ce1f" on path "vmhba38:C15:T1:L33" Failed: H:0x0 D:0x2 P:0x0 Valid sense data: 0x5 0x24 0x0. Act:NONE
2016-10-05T19:05:54.866Z cpu8:33291)NMP: nmp_ThrottleLogForDevice:3298: Cmd 0x12 (0x43a5cd77d5c0, 0) to dev "naa.6842b2b00064d53a0000074a4d22b445" on path "vmhba38:C0:T1:L1" Failed: H:0x0 D:0x2 P:0x0 Valid sense data: 0x5 0x24 0x0. Act:NONE
2016-10-05T19:05:54.868Z cpu11:33341)NMP: nmp_ThrottleLogForDevice:3298: Cmd 0x12 (0x43a5cd77d5c0, 0) to dev "naa.6842b2b00064d53a000038e54fe2ccfe" on path "vmhba38:C5:T1:L16" Failed: H:0x0 D:0x2 P:0x0 Valid sense data: 0x5 0x24 0x0. Act:NONE
2016-10-05T19:05:54.871Z cpu7:33013)NMP: nmp_ThrottleLogForDevice:3298: Cmd 0x12 (0x43a5cd77d5c0, 0) to dev "naa.6842b2b00064d53a000038f44fe2ce07" on path "vmhba38:C15:T1:L30" Failed: H:0x0 D:0x2 P:0x0 Valid sense data: 0x5 0x24 0x0. Act:NONE
2016-10-05T19:05:54.872Z cpu8:33291)NMP: nmp_ThrottleLogForDevice:3298: Cmd 0x12 (0x43a5cd77d5c0, 0) to dev "naa.6842b2b00064d53a00001dde4dfed20e" on path "vmhba38:C0:T1:L9" Failed: H:0x0 D:0x2 P:0x0 Valid sense data: 0x5 0x24 0x0. Act:NONE
2016-10-05T19:05:54.873Z cpu7:33013)NMP: nmp_ThrottleLogForDevice:3298: Cmd 0x12 (0x43a5cd77d5c0, 0) to dev "naa.6842b2b00064d53a000038f84fe2ce38" on path "vmhba38:C15:T1:L35" Failed: H:0x0 D:0x2 P:0x0 Valid sense data: 0x5 0x24 0x0. Act:NONE
2016-10-05T19:05:54.877Z cpu7:33284)NMP: nmp_ThrottleLogForDevice:3298: Cmd 0x12 (0x43a5cd77d5c0, 0) to dev "naa.6842b2b00070936a000036ea4fe2cc1d" on path "vmhba38:C31:T1:L23" Failed: H:0x0

På Console ser jeg med ALT+F2 og ALT+F12:

vmwarelangsom1

vmwarelangsom2

vmwarelangsom3

Løsning som forkortede genstart tiden helt ned til 8-9 minutter:

  1. Gå til VMware vSphere Client -> Vælg Server -> Configuration -> Storage Adapters
  2. Under ISCSI Software Adapter vælg vmhba38 -> højre klik og vælg Properties
    vmwarelangsom4
  3. Gå til Network Configuration -> Vælg alle tilføjede Port Group, f.eks. iSCSI01 og klik Remove.
    vmwarelangsom5
  4. Gå til Static Discovery -> Markere alle sammen og klik Remove -> Close
    vmwarelangsom6
  5. Klik Yes til Rescan af datastores.

 

Efter jeg genstartede VMware ESXi serveren, så kom ingen fejl eller disse msleep meddelser og hosten reboot tog ca. 9 minutter.

*Virtuelle servere oplevede frysning/delay for 2 sekunder, hvor de herefter fungerede fint.

Hardware:

196GB RAM, 24 Kerner
2 x ISCSi 1gbits netkort
2 x Management 1gbits netkort
Flash SD Kort, VMware ESXi 6.0
3 x DELL MD3200 SAN Storage

Skriv et svar

Din e-mailadresse vil ikke blive publiceret. Krævede felter er markeret med *