experchange > solaris

jaybraun2.0 (01-12-20, 11:52 PM)
I was told that Solaris 11 does not support 32-bit applications. I am curious as to whether this is possibly a mistaken impression, and perhaps also why this is the case.

Once can, of course, run Solaris 10 in a guest domain or a Solaris zone, and keep 32-bit applications.

The intent is to port all 32-bit apps to 64-bit, but my concern is one of timing/dependencies in migrating to Solaris 11.
Grant Taylor (01-13-20, 12:20 AM)
On 1/12/20 2:52 PM, jaybraun2.0 wrote:
> Once can, of course, run Solaris 10 in a guest domain or a Solaris
> zone, and keep 32-bit applications.


Won't the zone have the same Solaris 11 kernel, thus the same
limitations of said kernel?
Jay Braun (01-13-20, 01:16 AM)
On Sunday, January 12, 2020 at 2:20:59 PM UTC-8, Grant Taylor wrote:
> On 1/12/20 2:52 PM, jaybraun2.0 wrote:
> Won't the zone have the same Solaris 11 kernel, thus the same
> limitations of said kernel?
> --
> Grant. . . .
> unix || die


Interesting point, Grant, thank you. I am referring specifically to a recent procurement of SPARC T8 series servers. Would anyone know if the architecture precludes the execution of 32-bit applications, even on an Oracle 10 guest domain or Solaris zone?
Gary R. Schmidt (01-13-20, 04:55 AM)
On 13/01/2020 08:52, jaybraun2.0 wrote:
> I was told that Solaris 11 does not support 32-bit applications. I am curious as to whether this is possibly a mistaken impression, and perhaps also why this is the case.
> Once can, of course, run Solaris 10 in a guest domain or a Solaris zone, and keep 32-bit applications.
> The intent is to port all 32-bit apps to 64-bit, but my concern is one of timing/dependencies in migrating to Solaris 11. They work fine on my 11.1, 11.2, 11.3, and 11.4 SPARC[1] systems.


And Solaris Studio still defaults to producing 32-bit binaries. (That
may have changed with a more recent version.)

The kernel has been 64-bit only for quite a while, now.

Cheers,
Gary B-)

1 - And those compiled for 32-bit Intel work fine on the equivalent x64
systems.
YTC#1 (01-13-20, 10:39 AM)
On 12/01/2020 21:52, jaybraun2.0 wrote:
> I was told that Solaris 11 does not support 32-bit applications. I am curious as to whether this is possibly a mistaken impression, and perhaps also why this is the case.
> Once can, of course, run Solaris 10 in a guest domain or a Solaris zone, and keep 32-bit applications.
> The intent is to port all 32-bit apps to 64-bit, but my concern is one of timing/dependencies in migrating to Solaris 11.


Someone has been doling out some FUD.

Yes, you *can* run 32 bit apps on S11.x

Casper H.S. Dik (01-13-20, 11:38 AM)
jaybraun2.0 writes:

>I was told that Solaris 11 does not support 32-bit applications. I am curious as to whether this is possibly a mistaken impression, and perhaps also why this is the case.


This is incorrect.

Solaris 11 no longer has a support for 32 bit CPUs; it always runs
a 64 bit kernel and as a result you will need to have 64 bit kernel modules.

But 32 bit applications still run fine; in fact, most applications shipped
in Solaris 11/11 through 11.3 are mostly 32 bit.

Only in Solaris 11.4, the bulk of the applications were changed to 64 bit.

Casper
Jay Braun (01-13-20, 02:19 PM)
Thanks, everyone. I've implemented/led 64-bit ports on Linux for the sake of increased virtual address space, and we left some programs as 32-bit because address space was not a consideration for those applications. So, this claim seemed to have been a misconception -- or maybe I misunderstood theperson to whom I was talking. I'm going to assume no FUD was intended.

The link that Bruce provided mentioned the "Year 2038" issue that plagues all applications using 32-bit time with 1/1/1970 as the beginning of the epoch. But how are Solaris users dealing with the "Year 2034" issue of suspension of Oracle support? That might be a new topic, so please excuse the digression.
YTC#1 (01-14-20, 12:46 PM)
On 13/01/2020 12:19, Jay Braun wrote:
> Thanks, everyone. I've implemented/led 64-bit ports on Linux for the sake of increased virtual address space, and we left some programs as 32-bit because address space was not a consideration for those applications. So, this claim seemed to have been a misconception -- or maybe I misunderstood the person to whom I was talking. I'm going to assume no FUD was intended.
> The link that Bruce provided mentioned the "Year 2038" issue that plagues all applications using 32-bit time with 1/1/1970 as the beginning of the epoch. But how are Solaris users dealing with the "Year 2034" issue of suspension of Oracle support? That might be a new topic, so please excuse the digression.


I'll be retired (or dead) by then, so SEP (Somebody Else's Problem) :-)
Casper H.S. Dik (01-14-20, 05:22 PM)
YTC#1 <bdp> writes:

>On 13/01/2020 12:19, Jay Braun wrote:
>> Thanks, everyone. I've implemented/led 64-bit ports on Linux for the sake of increased virtual address space, and we left some programs as 32-bit because address space was not a consideration for those applications. So, this claim seemed to have been a misconception -- or maybe I misunderstood the person to whom I was talking. I'm going to assume no FUD was intended.
>> The link that Bruce provided mentioned the "Year 2038" issue that plagues all applications using 32-bit time with 1/1/1970 as the beginning of the epoch. But how are Solaris users dealing with the "Year 2034" issue of suspension of Oracle support? That might be a new topic, so please excuse the digression.


>I'll be retired (or dead) by then, so SEP (Somebody Else's Problem) :-)


Or $$$ as consultant?

Casper
YTC#1 (01-14-20, 06:37 PM)
On 14/01/2020 15:22, Casper H.S. Dik wrote:
> YTC#1 <bdp> writes:
>> On 13/01/2020 12:19, Jay Braun wrote:
>>> Thanks, everyone. I've implemented/led 64-bit ports on Linux for the sake of increased virtual address space, and we left some programs as 32-bit because address space was not a consideration for those applications. So, this claim seemed to have been a misconception -- or maybe I misunderstood the person to whom I was talking. I'm going to assume no FUD was intended.
>>> The link that Bruce provided mentioned the "Year 2038" issue that plagues all applications using 32-bit time with 1/1/1970 as the beginning of the epoch. But how are Solaris users dealing with the "Year 2034" issue of suspension of Oracle support? That might be a new topic, so please excuse the digression.

>> I'll be retired (or dead) by then, so SEP (Somebody Else's Problem) :-)

> Or $$$ as consultant?


I did well out of Y2K, but buy 2038 I will be 75. I really really would
hope to have stopped *well* before then :-)

I have recently considered brushing up on my Cobol though .... :-)
Similar Threads