Solaris Express 4/06 has been released. You can click here to download this release. There have been some minor bugfixes and some relatively noteworthy additions in this release (including PostgreSQL and PCIe support). Sun has a nice, detailed list of new features for this release.
Just as a FYI (forgot to mention in summary) this is based on build 36 of Nevada. That means the majority of the “goodies” in terms of ZFS fixes are there. For anybody running ZFS this is a must-have release.
There’s a ton more than just ZFS goodies. Check out my blog entry for details:
http://blogs.sun.com/roller/page/dp?entry=what_s_new_in_solaris13
Cool. Most of that was in the original summary link but I didn’t know about some of the others.
Honestly, your blog reads better than the “list of new features” link I posted from sun.com. Why don’t they use your blog as the basis for the release notes? +1 for you.
I saw ‘support for PCI Express’ – will that in turn result in better performance for PCI Express related hardware?
In reference to ZFS; whats the performance like?
“I saw ‘support for PCI Express’ – will that in turn result in better performance for PCI Express related hardware?”
I’m going to try and figure that out myself tonight. My system was unusable as a desktop with a 7900GT and Nvidia’s drivers with Sol 10U1. It stuttered like no other whenever there was any activity.
“In reference to ZFS; whats the performance like?”
Absolutely breathtaking. Even with Raid-Z. Check out:
http://www.opensolaris.org/jive/forum.jspa?forumID=80
Lots of good information there. I think these guys ( http://www.joyent.com/ ) use it on a 22TB array (going from memory..)
I’m going to try and figure that out myself tonight. My system was unusable as a desktop with a 7900GT and Nvidia’s drivers with Sol 10U1. It stuttered like no other whenever there was any activity.
Out of curiosity, was that using the Nvidia binary driver or using the status quo driver which comes with Xorg?
I guess I’m a little impatient about what I’d like from SUN; but having seen the progress, when Solaris 11 does come out, it will be a big leap from where it is now, in respects to server and workstation features.
To quote my post, “and Nvidia’s drivers”. Let me be clear though, this was Sol10U1, not Nevada.
Ah, ok, I assumed “Nvidia drivers” as in the bundled nvidia driver rather than just sticking to the default vesa driver which sometimes gets used by default.
Solaris is great… but installing takes forever. Is there a more user friendly installer in the Solaris Express releases? Specifically, I am hoping for something that will make it simpler to do advanced disk partitioning (raid, vm) and select packages that I need to install.
From OpenSolaris discussion lists, a (completely?) new design of installer is in the work, but may take long time.
Also, the whole ZFS thing (especially when it supports boot/root filesystems) will have a huge impact on user experience, including the install.
Today, I only do network install which is _much_ faster than CD/DVD install, but you do need another Solaris box.
What really strikes me on solaris (sunos 5.11.37) is how the mouse jumps in a jerky way whenever network funtion is called or disk operation is done; in both CDE and JDS environments.
The 2nd weired thing is that Nexenta alpha2 (a sunos+GNU”GNOME or KDE”) didn’t show this sign at all under the same hardware.
Well it is a development release you are bound to get regressions at times between builds. Perhaps if you have a mosey over to opensolaris.org and file a bug report.
Likewise when Nextenta update to use the same baseline as Nevada37 it be interesting to see if the issue shows up. (Alpha2 has been out for awhile correct?)
Nexenta (alpha2) uses solaris previous builds but with previous opensolaris that I have tested I noticed this weird behavior, and that’s why I tested this build to check weather they corrected the previous bugs; So I roughly concluded that the problem might be one of two:
1. Solaris kernel (not compatible and bug-fixed on x86 yet)
2. CDE or JDS environment problems that affect curser behavior.
And by subtracting Solaris kernel from the equation because it acts normal on nexenta then my final believe is that opensolaris suffers from the 2nd problem.
Then, why don’t they use GNU projects for their GUI?!
They cannot make better environment than GNOME/KDE,can they?
(and dropping sound and so on) from harddisk usage, I get those to. And it doesn’t have to be high load either, earlier (in b28?) when I where using a few torrents at the same time running at a few kB/s the whole desktop more or less stalled.
Nexanta uses a non-debug kernel, and maybe other differences aswell.
Dubhthach: The latest release of Nexanta is alpha 4.
We have discussed the choppy performance in #opensolaris but I haven’t got an answer I’m confident and comfortable with. The best suggestion I’ve heard is that it would be to “the x86 scheduler” or something, not that I know if the scheduler differs?
Or it might had been so easy that Sun prioritize i/o-performance over userface performance, but if that was the case someone would probably had said so. Please tell me if you figure out why it is this way.
This definitly sounds like interrupt related problems.
For example, something like this:
http://www.opensolaris.org/jive/thread.jspa?messageID=11873⹡
I can’t explain why Nexanta doesn’t have this problem, since it uses exactly the same kernel, of course the difference between debug and non-debug kernel can cause some different, I just can tell how and why.
But don’t worry, it should definitely be fixed. But you need to report it on OpenSolaris forum and push for the solution :-).
What kind of mobo/chipset do you have?
I ran into the same problems as this, and worked with seth to test stuff last week.
run
intrstat as root and see if you get a lot of interrupts on audio810 or some other device.
I have the same problem on Gigabyte 865PE chipset based with 2.8 GHz P4 CPU + HT enabled and with 1 GB DDR Dual channel, absolutely every OS I have tried on it was compatible (all Linuxes, Zeta, Nexenta, PC-BSD, freeBSD, skyOS,….). Jerky mouse happened with me only with solaris 5.11.x and Zeta 1.x . Zeta fixed the problem finally but solaris didn’t.
I can see the same behaviour, using a Sun Ultra 20.
Running intrstat(1m) doesn’t show anything, at least
during 3 to 4 sample tests.
The link in the release to the manpages notes points nowhere. I was curious if Sun mentioned, as do most sources, that the functions originated in OpenBSD. The point of interest here is that generally Sun does not do this.
(looking forward to reading the manpage).
The man pages points nowhere since docs.sun.com isn’t
aware of SX, SX:CR or ON. But you can browse the
source of OpenSolaris to see the corresponding source
files:
http://cvs.opensolaris.org/source/search?q=mkstemp+mkstemps+mkdtemp
Then you will see that the OpenBSD copyright is naturally
preserved, for example:
http://cvs.opensolaris.org/source/xref/on/usr/src/cmd/ssh/libopenbs…