vmware server -- disk performance

[sdino]

Hi,

ich habe auf einem Hetzner Rootserver:

CPU: AMD Athlon 64 X2 5600+ Dual Core
RAM: 4 GB DDR2 RAM
HD: 2xSATA II 400 GB, SAMSUNG_HD403LJ
MB: MICRO-STAR MS-7368

Debian Etch 4.0 64Bit und Vmware SRV 1.04 installiert.

Als Gast Win2003SRV R2.

Das Gastsystem ist sehr langsam bei Festplattenzugriffen. Das kopieren von 5 GB Daten innerhalb dieses Gasts dauert 60 Minuten.

Im vmware.log steht das:

Dec 21 07:13:59: vmx| SCSI0:0: Command READ(10) took 1.184 seconds (ok)
Dec 21 07:14:00: vmx| SCSI0:0: Command WRITE(10) took 1.286 seconds (ok)
Dec 21 07:14:00: vmx| SCSI0:0: Command WRITE(10) took 1.289 seconds (ok)
Dec 21 07:14:00: vmx| SCSI0:0: Command WRITE(10) took 1.196 seconds (ok)
Dec 21 07:14:58: vmx| SCSI0:0: Command WRITE(10) took 1.921 seconds (ok)
Dec 21 07:14:58: vmx| SCSI0:0: Command WRITE(10) took 1.744 seconds (ok)
Dec 21 07:14:58: vmx| SCSI0:0: Command WRITE(10) took 1.752 seconds (ok)
Dec 21 07:16:09: vmx| SCSI0:0: Command WRITE(10) took 1.052 seconds (ok)
Dec 21 07:16:09: vmx| SCSI0:0: Command WRITE(10) took 1.815 seconds (ok)
Dec 21 07:16:09: vmx| SCSI0:0: Command WRITE(10) took 1.851 seconds (ok)
Dec 21 07:16:09: vmx| SCSI0:0: Command WRITE(10) took 1.851 seconds (ok)
Dec 21 07:16:09: vmx| SCSI0:0: Command WRITE(10) took 1.806 seconds (ok)
Dec 21 07:16:09: vmx| SCSI0:0: Command WRITE(10) took 1.806 seconds (ok)
Dec 21 07:16:09: vmx| SCSI0:0: Command WRITE(10) took 1.745 seconds (ok)
Dec 21 07:16:09: vmx| SCSI0:0: Command WRITE(10) took 1.654 seconds (ok)
Dec 21 07:16:09: vmx| SCSI0:0: Command WRITE(10) took 1.607 seconds (ok)
Dec 21 07:16:09: vmx| SCSI0:0: Command WRITE(10) took 1.481 seconds (ok)
Dec 21 07:16:26: vmx| SCSI0:0: Command WRITE(10) took 1.114 seconds (ok)
Dec 21 07:16:26: vmx| SCSI0:0: Command WRITE(10) took 1.171 seconds (ok)
Dec 21 07:16:26: vmx| SCSI0:0: Command WRITE(10) took 1.187 seconds (ok)
Dec 21 07:16:26: vmx| SCSI0:0: Command WRITE(10) took 1.187 seconds (ok)
Dec 21 07:16:36: vmx| SCSI0:0: Command WRITE(10) took 1.177 seconds (ok)
Dec 21 07:16:36: vmx| SCSI0:0: Command WRITE(10) took 1.177 seconds (ok)
Dec 21 07:16:36: vmx| SCSI0:0: Command WRITE(10) took 1.177 seconds (ok)
Dec 21 07:16:36: vmx| SCSI0:0: Command WRITE(10) took 1.177 seconds (ok)
Dec 21 07:16:36: vmx| SCSI0:0: Command WRITE(10) took 1.132 seconds (ok)
Dec 21 07:16:36: vmx| SCSI0:0: Command WRITE(10) took 1.085 seconds (ok)
Dec 21 07:16:36: vmx| SCSI0:0: Command WRITE(10) took 1.034 seconds (ok)
Dec 21 07:16:44: vmx| SCSI0:0: Command WRITE(10) took 1.228 seconds (ok)
Dec 21 07:16:44: vmx| SCSI0:0: Command WRITE(10) took 1.268 seconds (ok)
Dec 21 07:16:44: vmx| SCSI0:0: Command WRITE(10) took 1.280 seconds (ok)
Dec 21 07:16:44: vmx| SCSI0:0: Command WRITE(10) took 1.180 seconds (ok)
Dec 21 07:16:44: vmx| SCSI0:0: Command WRITE(10) took 1.104 seconds (ok)
Dec 21 07:16:44: vmx| SCSI0:0: Command WRITE(10) took 1.110 seconds (ok)
Dec 21 07:16:49: vmx| SCSI0:0: Command WRITE(10) took 1.271 seconds (ok)
Dec 21 07:16:49: vmx| SCSI0:0: Command WRITE(10) took 1.355 seconds (ok)
Dec 21 07:16:49: vmx| SCSI0:0: Command WRITE(10) took 1.394 seconds (ok)
Dec 21 07:16:49: vmx| SCSI0:0: Command WRITE(10) took 1.482 seconds (ok)
Dec 21 07:16:49: vmx| SCSI0:0: Command WRITE(10) took 1.543 seconds (ok)
Dec 21 07:16:50: vmx| SCSI0:0: Command WRITE(10) took 2.204 seconds (ok)
Dec 21 07:16:50: vmx| SCSI0:0: Command WRITE(10) took 2.356 seconds (ok)
Dec 21 07:16:50: vmx| SCSI0:0: Command WRITE(10) took 2.356 seconds (ok)
Dec 21 07:16:50: vmx| SCSI0:0: Command WRITE(10) took 2.356 seconds (ok)
Dec 21 07:16:50: vmx| SCSI0:0: Command WRITE(10) took 1.735 seconds (ok)
Dec 21 07:16:50: vmx| SCSI0:0: Command WRITE(10) took 1.203 seconds (ok)
Dec 21 07:16:50: vmx| SCSI0:0: Command WRITE(10) took 1.001 seconds (ok)
Dec 21 07:16:57: vmx| SCSI0:0: Command WRITE(10) took 1.680 seconds (ok)
Dec 21 07:16:57: vmx| SCSI0:0: Command WRITE(10) took 1.839 seconds (ok)
Dec 21 07:16:57: vmx| SCSI0:0: Command WRITE(10) took 1.890 seconds (ok)
Dec 21 07:16:57: vmx| SCSI0:0: Command WRITE(10) took 2.040 seconds (ok)
Dec 21 07:16:57: vmx| SCSI0:0: Command WRITE(10) took 2.041 seconds (ok)
Dec 21 07:16:57: vmx| SCSI0:0: Command WRITE(10) took 2.024 seconds (ok)

Ist das normal?

[saxa]

Nein, es ist nicht normal.

Die SCSI-Timeouts weisen auf ein langsames oder nicht ordnungsgemäßes Storage-System des Hosts hin.

Wie sind diese 2 Platten formatiert? RAID?

[sdino]

Hallo,

nein, die 2 Platten sind nicht im Raid-Modus. Ganz normal SATA.

in Linux/Debian selber hab ich gute 40 MByte/Sec. Durchsatz beim Kopieren einer DAtei von HD1 auf HD2 mit Midnight Commander (mc) z.bsp.

Aber in Vmware ist dann alles grottenlangsam.

Any idea?

[MSueper]

Hi,
leg doch vorallokierte virtuelle Platten-Files an und achte darauf, das diese nicht zu sehr fragmentiert sind (sprich, das LW, warauf man sie anlegt sollte genug Platz haben).
Läuft bei Dir die VM im Logging-Modus? (= Werden redo-Files erstellt anstelle direkt in die vmdk zu schreiben?)
Martin

[sdino]

Hi,

was bedeutet vorallokiert? Dass man bei der Erstellung einer Hd den Speicherplatz definitiv auf der Host-HD reserviert und das File über die exakte Größe angelegt wird?
Wenn ja -- ich mach das immer so.

Bei diesen Servern (ich habe mehrere) habe ich ja immer 2 HD's:
- auf HD1 ist das Host OS und Vmware installiert
- HD2 sind nur die vmware-guests drauf

In diesem Fall ist auf HD2 eine Gast-Maschine, die Host-OS HD wurde vorher formattiert.

redo files find ich keine:

Code: Alles auswählen

-rw------- 1 root root 2.0G 2007-12-21 07:14 2GB-HDTACH-TEST-DISK-flat.vmdk
-rw------- 1 root root  362 2007-12-21 07:14 2GB-HDTACH-TEST-DISK.vmdk
-r--r--r-- 1 root root   11 2007-12-21 07:13 2GB-HDTACH-TEST-DISK.vmdk.WRITELOCK
-rw------- 1 root root 756M 2007-12-21 07:13 564d5677-2763-8673-9046-5b1db197ac87.vmem
-r--r--r-- 1 root root   11 2007-12-21 07:13 564d5677-2763-8673-9046-5b1db197ac87.vmem.WRITELOCK
-rw------- 1 root root 8.5K 2007-12-21 07:13 nvram
-rw-r--r-- 1 root root 180K 2007-12-21 07:11 vmware-0.log
-rw-r--r-- 1 root root 164K 2007-12-20 18:23 vmware-1.log
-rw-r--r-- 1 root root  25K 2007-12-20 17:05 vmware-2.log
-rw-r--r-- 1 root root 269K 2007-12-22 11:08 vmware.log
-rw------- 1 root root  30G 2007-12-22 19:13 WindowsServer2003StandardEdition-flat.vmdk
-rw------- 1 root root  376 2007-12-21 07:13 WindowsServer2003StandardEdition.vmdk
-r--r--r-- 1 root root   11 2007-12-21 07:13 WindowsServer2003StandardEdition.vmdk.WRITELOCK
-rw------- 1 root root    0 2007-12-20 13:05 Windows Server 2003 Standard Edition.vmsd
-rwxr-xr-x 1 root root 1.2K 2007-12-21 07:13 Windows Server 2003 Standard Edition.vmx


Ich habe jetzt 5 solche SRV, alle bei Hetzner, verschiedene HW, die Gast-Disk Operationen sind auf allen Servern sehr schlecht.

Ich hab da wohl ein grundlegendes Problem?

[sdino]

Hi,

dürften auf einem normalen System solche Meldungen:

Code: Alles auswählen

Dec 21 07:16:50: vmx| SCSI0:0: Command WRITE(10) took 2.356 seconds (ok)
Dec 21 07:16:50: vmx| SCSI0:0: Command WRITE(10) took 1.735 seconds (ok)
Dec 21 07:16:50: vmx| SCSI0:0: Command WRITE(10) took 1.203 seconds (ok)
Dec 21 07:16:50: vmx| SCSI0:0: Command WRITE(10) took 1.001 seconds (ok)
Dec 21 07:16:57: vmx| SCSI0:0: Command WRITE(10) took 1.680 seconds (ok)
Dec 21 07:16:57: vmx| SCSI0:0: Command WRITE(10) took 1.839 seconds (ok)


nicht auftreten?