Re: [Fink-devel] postgresql80-unified.patch

2006-02-02 Thread Benjamin Reed
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Benjamin Reed wrote: Ah, so they changed the kernel so that settings *can* be changed later? It used to be once you set the shm* values, they were stuck, so it had to be in sysctl.conf so they got set first. ...and if that's the case, can't we

Re: [Fink-devel] postgresql80-unified.patch

2006-02-02 Thread Neil Tiffin
This is directly from my /etc/rc file if [ -f /etc/sysctl.conf ]; then awk '{ if (!index($1, #) index($1, =)) print $1 }' /etc/ sysctl.conf | while read do sysctl -w ${REPLY} done fi #sysctl -w kern.sysv.shmmax=4194304 kern.sysv.shmmin=1

Re: [Fink-devel] postgresql80-unified.patch

2006-02-02 Thread Benjamin Reed
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Neil Tiffin wrote: This is directly from my /etc/rc file if [ -f /etc/sysctl.conf ]; then awk '{ if (!index($1, #) index($1, =)) print $1 }' /etc/sysctl.conf | while read do sysctl -w ${REPLY} done fi #sysctl -w

Re: [Fink-devel] postgresql80-unified.patch

2006-02-02 Thread Benjamin Reed
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Martin Costabel wrote: As you noticed, on Tiger (and on recent Panthers) this won't work. No matter what values for kern.sysv.shm* you write into /etc/sysctl.conf, they will not stick, because they are reset in /etc/rc afterwards. Ah, so they

Re: [Fink-devel] postgresql80-unified.patch

2006-02-02 Thread Martin Costabel
Benjamin Reed wrote: [] as far as I'm aware, on OSX the only time that the values actually get set is the *first time*, after bootup, that you run, say: sysctl -w kern.sysv.shmmax=8388608 So if you have a sysctl.conf, it will have already set them. The line you've commented out can't change

Re: [Fink-devel] postgresql80-unified.patch

2006-02-02 Thread Neil Tiffin
Yes, I just tried it again and it worked as you stated. Only retains the first value set. I spent considerable time trying to figure this out and the only solution that worked was to edit /etc/rc and reboot. I even saw it change the settings the second time on the boot screen. That is

Re: [Fink-devel] postgresql80-unified.patch

2006-01-28 Thread Neil Tiffin
What I actually asked for was to have the suggested shm* values added to the failure message if you already have an /etc/sysctl.conf. This way the user has some idea was is needed when it does not work and can adjust accordingly. This still should be done. After thinking about it, I also

Re: [Fink-devel] postgresql80-unified.patch

2006-01-28 Thread Martin Costabel
Neil Tiffin wrote: [] one. In adiition, if writing a /etc/sysctl.conf will not work (and that seems to be the consensus) then fink should write a message stating this. Yes, I agree. I am convinced this whole section of the postgresql package with its allusion to /etc/sysctl.conf is wrong. If

Re: Fink feedback for neilt (Re: [Fink-devel] postgresql80-unified.patch)

2006-01-28 Thread Benjamin Reed
On 1/28/06, Martin Costabel [EMAIL PROTECTED] wrote: Yes, I agree. I am convinced this whole section of the postgresql package with its allusion to /etc/sysctl.conf is wrong. If the higher values of the shm stuff are really needed, then the package should give instructions to the user how to

[Fink-devel] postgresql80-unified.patch

2006-01-26 Thread Neil Tiffin
I would like to see this changed to this + if [ -f /etc/sysctl.conf ]; then + echoYou already have an /etc/sysctl.conf so I'm assuming you know what you're + echodoing. Please update your shared memory settings to something higher and + echo

Re: [Fink-devel] postgresql80-unified.patch

2006-01-26 Thread Martin Costabel
Neil Tiffin wrote: I would like to see this changed to this As you noticed, on Tiger (and on recent Panthers) this won't work. No matter what values for kern.sysv.shm* you write into /etc/sysctl.conf, they will not stick, because they are reset in /etc/rc afterwards. The only way to