[rescue] Solaris not stable till release 2.x?????? WAS::Re: [geeks] Rumors flying
mcguire at neurotica.com
Tue Dec 6 18:05:37 CST 2016
2.3 ran in short spurts between crashes. 2.4 ran in longer spurts
between crashes. During a good time, 2.4 would stay up for about two
whole weeks. Contrast that with the SunOS4 machines on the other side
of the datacenter which frequently had uptimes measured in multiple years.
The typical failure mode was kernel panics (usually in the network
stack but sometimes in device drivers) and hard lockups. There were no
hardware problems. (i.e., same behavior on 100+ machines)
When they were running, the 2.3 and 2.4 machines, running on
Ultra1-class hardware, would handle about half of the workload that a
high-config SPARCstation-20 running SunOS4 would. This made Solaris2
rather unpopular in our environment even before the stability problems
became apparent. The term "slower than pissing tar" was often heard.
These were high-workload, forward-facing network servers in a
large-scale ISP environment.
On 12/06/2016 06:38 PM, Jerry Kemp wrote:
> I'm kind of curious regarding comments of Solaris wasn't stable till
> release 2.? , where the ? is the particular version you choose.
> A bit of my background. I got started with Solaris during my time in
> the military with Solaris 2.4, which, from my recollection, shipped with
> enough bugs that in addition to the Solaris 2.4 CD, it shipped with a
> separate CD full of patches. Solaris (SPARC) boxes were put into place
> to replace aging ATT 3B2 600G & 600GR systems.
> For our needs, those 2.4 systems ran without issue, at least anything
> specific to the OS, till they were upgraded to 2.5 then 2.5.1.
> A few years down the road, I left the military, and found myself in a
> large oil company with responsibility of administering 5-600 Sparc
> servers and workstations, along with a handful of SGI Irix boxes and a
> one off AIX box that ran some simulation I never really understood.
> Several floors of geologist, each having a Sparc based workstation on
> their desk, along with a PC/desktop to read Novell/Groupware email.
> At that location I had a handful of small servers running Solaris 2.3,
> that took several years to replace/upgrade. My request to upgrade the
> 2.3 boxes fell on deaf ears as they continually ran without issue.
> The first big problem I did encounter (same oil company) was on some big
> file servers sharing out about 2 Tb of data, at least that (2 Tb) seemed
> large for the late 1990's. The data was stored on Veritas foundation
> suite FS, and when these boxes were upgraded from 2.5.1 to 2.6, no one
> had checked with Veritas to check for compatibility.
> It turns out that there were big kernel changes between 2.5.1 and 2.6.
> But the problem was really a human one vs either a problem with Solaris
> or Veritas. There has been a few Solaris releases where I felt Sun went
> Mac OS X on us, AKA, the new release took away more features and support
> then providing new stuff. But, from 2.3 to present, baring hardware
> problems, Solaris has generally proven itself to be rock solid.
> I realize that there are numerous issues, like when Solaris 2.0 came out
> and took away the BSD rc start up scripts to give us the Sys V scripts.
> And similar things happened with Solaris 10 with SMF, although the Sys V
> start up scripts are still with us but depreciated. These were
> unpopular decisions.
> Closing out by repeating my original question, I'm curious regarding
> comments of Solaris wasn't stable till release 2.? , where the ? is the
> particular version you choose. What was so broken that you didn't
> consider the Solaris stable at that point?
> On 12/ 6/16 04:58 PM, Stefan Skoglund (lokal anvC$ndare) wrote:
>>> sparky:~$ uname -a
>>> SunOS sparky 5.5.1 Generic_103640-27 sun4c sparc SUNW,Sun_4_75
>>> sparky:~$ uptime
>>> 4:36pm up 77 day(s), 23:41, 1 user, load average: 0.06, 0.02, 0.02
>>> Gotta love that old 4c iron.
>> Yes, i remember. 2.5.1 was the first one which was good enough if it
>> wasn't awfully important to run SunOS 2.x for some reason or another.
>> The workstations had a habit of hanging once in a month or so.
>> rescue list - http://www.sunhelp.org/mailman/listinfo/rescue
> rescue list - http://www.sunhelp.org/mailman/listinfo/rescue
Dave McGuire, AK4HZ
New Kensington, PA
More information about the rescue