Thread
-
Re: "--tuning" compile and runtime option (?)
John Gray <jgray@beansindustry.co.uk> — 2001-04-10T09:52:25Z
I can understand why there might be some resistance to the idea of adding performance tuning flags into the server rather than documenting existing settings better, but I think a compromise would be possible. Could we develop a helper application that takes the --tuning superserver argument and translates that into a set of options to pass? That way, fine-tuning by hand is still practical, but for those who just want a good first set of values, a tuning helper application that looks at system memory, processor speed, and a user-supplied indication of the system's purpose and produces a set of postmaster options might be the way to approach this. And we don't bloat the server with extra algorithms. There is no dependency on this utility, either (assuming we continue to use workable defaults for postmaster options!) but it may benefit some people to use it. I really like the performance hints thing too. John -- John Gray Tel +44-7974-100-584 mailto:jgray@beansindustry.co.uk
-
Re: Re: "--tuning" compile and runtime option (?)
Bruce Momjian <pgman@candle.pha.pa.us> — 2001-04-10T14:37:05Z
The problem is that I can't figure out what would be tuned by these options. We only have 2-3 parameters that can be changed. > I can understand why there might be some resistance to the idea of > adding performance tuning flags into the server rather than documenting > existing settings better, but I think a compromise would be possible. > > Could we develop a helper application that takes the --tuning > superserver argument and translates that into a set of options to pass? > That way, fine-tuning by hand is still practical, but for those who just > want a good first set of values, a tuning helper application that looks > at system memory, processor speed, and a user-supplied indication of the > system's purpose and produces a set of postmaster options might be the > way to approach this. And we don't bloat the server with extra > algorithms. There is no dependency on this utility, either (assuming we > continue to use workable defaults for postmaster options!) but it may > benefit some people to use it. > > I really like the performance hints thing too. > > John > > -- > John Gray > Tel +44-7974-100-584 > mailto:jgray@beansindustry.co.uk > > > > ---------------------------(end of broadcast)--------------------------- > TIP 5: Have you checked our extensive FAQ? > > http://www.postgresql.org/users-lounge/docs/faq.html > -- Bruce Momjian | http://candle.pha.pa.us pgman@candle.pha.pa.us | (610) 853-3000 + If your life is a hard drive, | 830 Blythe Avenue + Christ can be your backup. | Drexel Hill, Pennsylvania 19026
-
Re: Re: "--tuning" compile and runtime option (?)
Mitch Vincent <mitch@venux.net> — 2001-04-10T14:47:13Z
Indeed, as an avid user (and tuner, I suppose) of PostgreSQL, I can't see how any configure option would be faster or better than the existing command line /config file parameters -- it would only serve to make things harder to deal with IMHO. "Tuning" PostgreSQL is pretty simple, and is explained pretty well throughout the manual (especially in the section titled "Understanding Performance"). We have -S -B and the fsync options, <Austin Powers Voice> That's about it.. </Austin Powers Voice> --- right? All are explained in the manual and are as easy to use as anyone could ask... Any OS tuning should be left up to the administrator as that's what administrators are for :-) Just my humble $0.02 worth.. -Mitch Software development : You can have it cheap, fast or working. Choose two. ----- Original Message ----- From: "Bruce Momjian" <pgman@candle.pha.pa.us> To: "John Gray" <jgray@beansindustry.co.uk> Cc: <pgsql-hackers@postgresql.org> Sent: Tuesday, April 10, 2001 10:37 AM Subject: Re: Re: "--tuning" compile and runtime option (?) > > The problem is that I can't figure out what would be tuned by these > options. We only have 2-3 parameters that can be changed. > > > > I can understand why there might be some resistance to the idea of > > adding performance tuning flags into the server rather than documenting > > existing settings better, but I think a compromise would be possible. > > > > Could we develop a helper application that takes the --tuning > > superserver argument and translates that into a set of options to pass? > > That way, fine-tuning by hand is still practical, but for those who just > > want a good first set of values, a tuning helper application that looks > > at system memory, processor speed, and a user-supplied indication of the > > system's purpose and produces a set of postmaster options might be the > > way to approach this. And we don't bloat the server with extra > > algorithms. There is no dependency on this utility, either (assuming we > > continue to use workable defaults for postmaster options!) but it may > > benefit some people to use it. > > > > I really like the performance hints thing too. > > > > John > > > > -- > > John Gray > > Tel +44-7974-100-584 > > mailto:jgray@beansindustry.co.uk > > > > > > > > ---------------------------(end of broadcast)--------------------------- > > TIP 5: Have you checked our extensive FAQ? > > > > http://www.postgresql.org/users-lounge/docs/faq.html > > > > > -- > Bruce Momjian | http://candle.pha.pa.us > pgman@candle.pha.pa.us | (610) 853-3000 > + If your life is a hard drive, | 830 Blythe Avenue > + Christ can be your backup. | Drexel Hill, Pennsylvania 19026 > > ---------------------------(end of broadcast)--------------------------- > TIP 2: you can get off all lists at once with the unregister command > (send "unregister YourEmailAddressHere" to majordomo@postgresql.org) > -
RE: Re: Re: "--tuning" compile and runtime option (?)
Darren King <darrenk@insightdist.com> — 2001-04-10T15:13:19Z
> I can't see how any configure option would be faster or > better than the existing command line /config file parameters > -- it would only serve to make things harder to deal with IMHO. > "Tuning" PostgreSQL is pretty simple, and is explained pretty > well throughout the manual (especially in the section titled > "Understanding Performance"). We have -S -B and the fsync > options, <Austin Powers Voice> That's about it.. </Austin Powers > Voice> --- right? > All are explained in the manual and are as easy to use as anyone could > ask... Any OS tuning should be left up to the administrator > as that's what administrators are for :-) Someday, maybe other parameters (such as some of the optimizer's default values) will be configurable/changable on the fly. Would seem silly to have to stop/start the server to change those. Just an example, I'm sure there are other internal values that admins would like to tweak without recompiling or restarting. Would be much nicer to login as the superuser to do it rather than shut down my site. darrenk
-
Re: Re: Re: "--tuning" compile and runtime option (?)
Mitch Vincent <mitch@venux.net> — 2001-04-10T15:26:02Z
Well, what is being discussed here would require a *recompile* to change the values, so you sure wouldn't want that! :-) Currently you can set a lot of internal variables with the SET command in any PG session.. It's useful for forcing index or sequential scans and such (and I'm sure a lot more).. -Mitch Software development : You can have it cheap, fast or working. Choose two. ----- Original Message ----- From: "Darren King" <darrenk@insightdist.com> To: <pgsql-hackers@postgresql.org> Sent: Tuesday, April 10, 2001 11:13 AM Subject: RE: Re: Re: "--tuning" compile and runtime option (?) > > I can't see how any configure option would be faster or > > better than the existing command line /config file parameters > > -- it would only serve to make things harder to deal with IMHO. > > "Tuning" PostgreSQL is pretty simple, and is explained pretty > > well throughout the manual (especially in the section titled > > "Understanding Performance"). We have -S -B and the fsync > > options, <Austin Powers Voice> That's about it.. </Austin Powers > > Voice> --- right? > > All are explained in the manual and are as easy to use as anyone could > > ask... Any OS tuning should be left up to the administrator > > as that's what administrators are for :-) > > Someday, maybe other parameters (such as some of the optimizer's default > values) will be configurable/changable on the fly. Would seem silly to > have to stop/start the server to change those. > > Just an example, I'm sure there are other internal values that admins > would like to tweak without recompiling or restarting. Would be much > nicer to login as the superuser to do it rather than shut down my site. > > darrenk > > > ---------------------------(end of broadcast)--------------------------- > TIP 2: you can get off all lists at once with the unregister command > (send "unregister YourEmailAddressHere" to majordomo@postgresql.org) >
-
Re: Re: Re: "--tuning" compile and runtime option (?)
Bruce Momjian <pgman@candle.pha.pa.us> — 2001-04-10T15:35:19Z
We do have postgresql.conf. I am unsure how I would tune those based on a single flag. > > I can't see how any configure option would be faster or > > better than the existing command line /config file parameters > > -- it would only serve to make things harder to deal with IMHO. > > "Tuning" PostgreSQL is pretty simple, and is explained pretty > > well throughout the manual (especially in the section titled > > "Understanding Performance"). We have -S -B and the fsync > > options, <Austin Powers Voice> That's about it.. </Austin Powers > > Voice> --- right? > > All are explained in the manual and are as easy to use as anyone could > > ask... Any OS tuning should be left up to the administrator > > as that's what administrators are for :-) > > Someday, maybe other parameters (such as some of the optimizer's default > values) will be configurable/changable on the fly. Would seem silly to > have to stop/start the server to change those. > > Just an example, I'm sure there are other internal values that admins > would like to tweak without recompiling or restarting. Would be much > nicer to login as the superuser to do it rather than shut down my site. > > darrenk > > > ---------------------------(end of broadcast)--------------------------- > TIP 2: you can get off all lists at once with the unregister command > (send "unregister YourEmailAddressHere" to majordomo@postgresql.org) > -- Bruce Momjian | http://candle.pha.pa.us pgman@candle.pha.pa.us | (610) 853-3000 + If your life is a hard drive, | 830 Blythe Avenue + Christ can be your backup. | Drexel Hill, Pennsylvania 19026