Re: [AOLSERVER] Solaris 2.5.1 -- My God.
Without having any real metrics, the performance jump from Solaris 2.6 to 7 has been enormous. It's also much better in terms of the separation of the thread library (which is why Solaris 7 binaries don't run on Solaris 2.6) and sundry LWP changes that are not insignificant. You should be able to upgrade your older computer--Solaris 7 and 8 still run well on sun4m (a.k.a. SuperSparc, inside SparcStation 10s and 20s) but you'll need to adjust the AOLserver include/Makefile.global to use something other than "ultrasparc" and "sparcv9" because everything we touch Sun-wise is sun4u and better. Kris
Re: [AOLSERVER] Solaris 2.5.1 -- My God.
Jeff Hobbs wrote: [stuff] >Kriston Rehberg wrote: > [stuff] >I might add that you will have a hard time getting a compiler and/or >support for 2.5.1 from Sun nowadays. They want this to disappear, and >all I can say is that they have good reason. People should be on the >2.6+ for better stability, improved performance, ... > Yes, yes, thank you all, but in the past two months I have indeed moved to another server entirely. Which (in fact) I mentioned in passing last month. But I'm gonna miss that ol' Sparc. It was a handy workhorse for a long time. I'm thinking it might be happy as a home file server. (After moving most of its functionality to a dual-Pentium Dell I have no desire to use the Sparc as a server any more.)
Re: [AOLSERVER] Solaris 2.5.1 -- My God.
Kriston Rehberg wrote: > On Wed, 6 Feb 2002 02:28:28 -0500, Michael Roberts > <[EMAIL PROTECTED]> wrote: > > >...the AOL team uses Solaris 2.6 as a test environment. The burning question > >uppermost in my mind (save one) is HOW DIFFERENT COULD IT BE?!?!?!? > > I can't really say what is different about it except for some truly major > pthread/LWP changes, something to do with POSIX signalling, and the space of > about three years. I just gave away my remaining Solaris 2.6 dev box so > Solaris 7 is the low water mark now and I can't think of anything to say > except a pithy "upgrade your operating system" remark. If you can, get > gcc-2.95.3 and compile using -fPIC (gmake gccme=1) and pray for the best. > Even better, download the recommended patch roll-up for 2.5.1, download the > evaluation copy of the Sun C compiler and its requisite 2.5.1 patches, and > start over, possibly doing "gmake nativeme=1". I might add that you will have a hard time getting a compiler and/or support for 2.5.1 from Sun nowadays. They want this to disappear, and all I can say is that they have good reason. People should be on the 2.6+ for better stability, improved performance, ... Jeff
Re: [AOLSERVER] Solaris 2.5.1 -- My God.
On Wed, 6 Feb 2002 02:28:28 -0500, Michael Roberts <[EMAIL PROTECTED]> wrote: >...the AOL team uses Solaris 2.6 as a test environment. The burning question >uppermost in my mind (save one) is HOW DIFFERENT COULD IT BE?!?!?!? I can't really say what is different about it except for some truly major pthread/LWP changes, something to do with POSIX signalling, and the space of about three years. I just gave away my remaining Solaris 2.6 dev box so Solaris 7 is the low water mark now and I can't think of anything to say except a pithy "upgrade your operating system" remark. If you can, get gcc-2.95.3 and compile using -fPIC (gmake gccme=1) and pray for the best. Even better, download the recommended patch roll-up for 2.5.1, download the evaluation copy of the Sun C compiler and its requisite 2.5.1 patches, and start over, possibly doing "gmake nativeme=1". Regards, Kris
Re: [AOLSERVER] Solaris 2.5.1 -- My God.
Well, I figured this one out, actually -- after roughly eight hours of reading the entire Internet, I discovered that -Wl,-E is required as a ld flag in order to export all the symbols in the main program so that nssock.so can use them. ... Is this obsolete 2.5.1 behavior that Solaris 2.6 has dropped? BUT now that nssock and everybody loads fine, the server gets as far as the first idle, upon which it dies ignominiously, which event my good friend truss thoughtfully records: 24485: Incurred fault #6, FLTBOUNDS %pc = 0x00AA18A8 24485:siginfo: SIGSEGV SEGV_MAPERR addr=0x00AA18A8 24485: Received signal #11, SIGSEGV [caught] 24485:siginfo: SIGSEGV SEGV_MAPERR addr=0x00AA18A8 24485: sigprocmask(SIG_SETMASK, 0xEF6C36E4, 0x) = 0 24485: sigaction(SIGSEGV, 0xE140, 0x) = 0 24485: setcontext(0xE280) 24485: Incurred fault #6, FLTBOUNDS %pc = 0x00AA18A8 24485:siginfo: SIGSEGV SEGV_MAPERR addr=0x00AA18A8 24485: Received signal #11, SIGSEGV [default] 24485:siginfo: SIGSEGV SEGV_MAPERR addr=0x00AA18A8 Now -- what I can't figure out -- according to the project home, the AOL team uses Solaris 2.6 as a test environment. The burning question uppermost in my mind (save one) is HOW DIFFERENT COULD IT BE?!?!?!? (The burning question *truly* uppermost in my mind is of course "how can I extract myself from this situation" and I would greatly appreciate any help at all.) Michael Michael Roberts wrote: > I've compiled AS3.4.2 on Solaris 2.5.1 successfully, but upon attempting > to start it, I'm getting a relocation error: > > Warning: modload: failed to load '/usr/local/aolserver/bin/nssock.so': > 'ld.so.1: bin/nsd76: fatal: relocation error: symbol not found: > Ns_RegisterDriver: referenced in /usr/local/aolserver/bin/nssock.so'
[AOLSERVER] Solaris 2.5.1
I've compiled AS3.4.2 on Solaris 2.5.1 successfully, but upon attempting to start it, I'm getting a relocation error: Warning: modload: failed to load '/usr/local/aolserver/bin/nssock.so': 'ld.so.1: bin/nsd76: fatal: relocation error: symbol not found: Ns_RegisterDriver: referenced in /usr/local/aolserver/bin/nssock.so' (Not just nsd76, either -- I figured it couldn't hurt to try both.) What gives? What could I possibly be doing wrong? Michael
