Looking over the fence

I don’t know much about solaris in general so I thought i’d have a look at the upcoming release of opensolaris .  Being inquisitive can often get you into trouble

Anyway,  I fearlessly downloaded the Preview 2 iso and fired it up in vmware-server.  It’s no where near as slick looking as say an OpenSuSE Linux install, but it’s certainly come quite a way since I last looked at it.   It’s a Live CD environment that lets you have a bit of a look before deciding to install – as this was a vmware guest I just went ahead and installed it.

The install went through pretty cleanly, except there was no stinking network detected and of course we all know we must have a stinking network.

I eventually discovered that you need to tell vmware to emulate an e1000 device rather than the usual pcnet32 style device i’d come to know and love in linux.

To do this, just crack open your favourite editor called vi and edit your guest vmx file and add

ethernet0.virtualDev = “e1000″

The network card should be automagically detected on the next reboot and order will return to the universe.

Now that we have a network I can actually check to see what this puppy can do and i’m sure I can waste some more of this elusive thing called ’spare time’ – hey at least I wasn’t logged onto work doing crap paperwork so that’s got to be a good thing

Gutsy and Vmware-Server – you’re in for some pain

For those who are looking to run vmware-server on gutsy there are some things you really need to do to the vmware-server config after you’ve installed it otherwise you will get some seriously crap performance.

The 20 cent summary is if you make the following changes to the vmware-server config file

prefvmx.useRecommendedLockedMemSize = “TRUE”
prefvmx.minVmMemPct = “100″

and then add the following to each of your vm’s vmx files then performance will return to something quite acceptable.

sched.mem.pshare.enable = “FALSE”
mainMem.useNamedFile = “FALSE”
MemTrimRate = “0″
MemAllowAutoScaleDown = “FALSE”

Wonder about the level of performance improvement? Surely it can’t be that big a thing? Here is some sar output where the *only* changes made to the environment were the above changes.

15:25:01        CPU     %user     %nice   %system   %iowait    %steal     %idle
18:25:01        all     11.73      0.00     60.93      0.32      0.00     27.02
18:35:01        all     12.18      0.00     55.72      0.73      0.00     31.37
18:45:01        all     10.93      0.00     58.32      1.03      0.00     29.72
18:55:01        all     12.23      0.00     48.99      0.72      0.00     38.06
19:05:01        all     11.09      0.00     64.69      0.46      0.00     23.76
19:15:01        all     12.24      0.00     55.00      0.74      0.00     32.03
19:25:01        all     11.83      0.00     51.72      0.73      0.00     35.71
19:35:01        all     12.22      0.00     58.18      0.52      0.00     29.07
19:45:01        all     12.81      0.00     50.40      0.62      0.00     36.16
19:55:01        all     10.31      0.00     62.12      3.46      0.00     24.11
20:05:01        all      1.00      0.00     11.85      6.60      0.00     80.55
20:15:01        all      0.03      0.00      0.48      0.40      0.00     99.09
20:25:01        all      0.01      0.00      0.48      0.35      0.00     99.16
20:35:01        all      0.01      0.00      0.41      0.24      0.00     99.34
20:45:01        all      0.02      0.00      0.46      0.20      0.00     99.32
20:55:01        all      0.01      0.00      0.38      0.20      0.00     99.40
21:05:01        all      0.01      0.00      0.39      0.26      0.00     99.34
21:15:01        all      0.02      0.00      0.39      0.37      0.00     99.21
21:25:01        all      0.01      0.00      0.51      0.30      0.00     99.18
21:35:01        all      0.01      0.00      0.37      0.23      0.00     99.38
21:45:01        all      0.02      0.00      0.43      0.22      0.00     99.33
21:55:01        all      0.00      0.00      0.28      0.22      0.00     99.49

Thanks Cath for finding all that!