Thread: Search machine is ready
Somebody get a hold of me so we can do username/passwords etc... -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/
Dave Page wrote: > I am forwarding this to pgsql-www as it really is a community decision. > > On 14/2/06 18:00, "Joshua D. Drake" <jd@commandprompt.com> wrote: > >>> To setup user accounts, to install/upgrade PostgreSQL/Aspseek/apache, to >> Well you shouldn't need root access for any of that. Nor should you be >> upgrading anything but Aspseek. If we need something upgrading we will >> gladly handle it. >> >>> fix things when they go wrong... We will continue to run and manage >>> things on a day to day basis, just with Command Prompt as backup. >>> >>> >>> I thought this was a dedicated machine? >> It is a dedicated machine. We don't allow root access to anyone on any >> of the machines we host, regardless if it is dedicated. >> >> This way we don't get weird things like people not knowing "who" >> installed an alpha version of mailman (like on pgFoundry). >> >> Our experience shows that if you think you need root access, chances >> are you don't. From your above requirements, you don't need root access. >> >> You do need some group perms setup, sudo ability to a couple of >> administrative commands. >> >> I know this is a little different then how the community is used to >> working but we have a solid track record and this is one of the reasons. > > Hi Joshua, > > You are right that what you are proposing is not the way we are used to > working, and having consulted with the other admins in the community > (specifically Marc and Magnus), we agree that we do not want to change the > way we work and have to rely on people that we don't know, and have no > intimate knowledge of our requirements or systems in a time when we might > need almost instant response in the middle of the night. > > We would be prepared to compromise and be restricted to root control of a VM > if you're prepared to setup the box as a Linux Vserver however. That would > ensure that only your staff have control over the host OS, allowing them to > support the system and network etc, but give us the control we require over > the software we run. > > Regards, Dave Well John Hansen didn't have a problem with it and has already begun work. I have also already stated that we can make sure that you actually have access to what you "need" to operate the environment. Your requirements as listed above distinctly note a lack of a need for root access. Thus I really don't know what your concern is outside of a feeling that you need "control". Sincerely, Joshua D. Drake -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/
On 14/2/06 21:38, "Joshua D. Drake" <jd@commandprompt.com> wrote: > Dave Page wrote: >> > > I am forwarding this to pgsql-www as it really is a community decision. Yes, which is why I discussed it with the community admins before responding to you. > Well John Hansen didn't have a problem with it and has already begun > work. I have also already stated that we can make sure that you actually > have access to what you "need" to operate the environment. > > Your requirements as listed above distinctly note a lack of a need for > root access. Thus I really don't know what your concern is outside of a > feeling that you need "control". Yes, we do. The community should control things, not individual companies. None of the other entities providing the project with hosting have gone out of their way to deny us the ability to control the services we provide, and as the people who ultimately have to make sure these things work we do not feel that having to rely on (for example) a level 1 support tech in the middle of the night who has no idea what our server does or how it works is in any way a good thing. This is a dedicated server for the PostgreSQL project, that you offered to us knowing full well we were expecting full root access to. Why is it such a bad thing to provide exactly that? Regards, Dave.
Dave Page wrote: > > > On 14/2/06 21:38, "Joshua D. Drake" <jd@commandprompt.com> wrote: > >> Dave Page wrote: >> I am forwarding this to pgsql-www as it really is a community decision. > > Yes, which is why I discussed it with the community admins before responding > to you. > >> Well John Hansen didn't have a problem with it and has already begun >> work. I have also already stated that we can make sure that you actually >> have access to what you "need" to operate the environment. >> >> Your requirements as listed above distinctly note a lack of a need for >> root access. Thus I really don't know what your concern is outside of a >> feeling that you need "control". > > Yes, we do. The community should control things, not individual companies. Are you saying that Command Prompt is not a part of the community? Or myself for that matter? > None of the other entities providing the project with hosting have gone out > of their way to deny us the ability to control the services we provide, I am not going out of my way. I am not sure why you have such a problem with this. 1. This is the way that Command Prompt, Inc. does hosting for all people that it hosts. Nobody gets root access. 2. You still haven't provided a single reason why you "need" root access. > as the people who ultimately have to make sure these things work we do not > feel that having to rely on (for example) a level 1 support tech in the > middle of the night who has no idea what our server does or how it works is > in any way a good thing. What level 1 support tech would this be? I don't have any level 1 support techs. Remember, we are not a hosting company. > > This is a dedicated server for the PostgreSQL project, that you offered to > us knowing full well we were expecting full root access to. Well no. I didn't expect that you would need root access because I have 50 dedicated machines at the facility none of which need root access. I am not trying to be difficult here but all I see is, "Well we like to do things as root and since this is only a community machine you should let us." I on the other hand am trying to bring a certain level of stability and quality to the infrastructure. That requires a level of discipline which means we use things like sudo, acls and group rights. We don't use root. > Why is it such a > bad thing to provide exactly that? Dave, come on. You are smarter then that. System Administrator 101 says you don't use root unless you have too. You don't give root to those who don't need it. I refer to #2 above. Nobody has given me one task that you would need to require root on this machine. You will need to configure apache... I will make sure you can do so via included confs. You will need to configure Aspseek... You will be able to do so. You will need to be able to stop/start postgresql. You will be able to do so via sudo. You will need to be able stop/start apache... You will be able to do so via sudo. You will need to be able to add users... I can give you sudo rights to do that (although that scares me a bit). You want to be able to upgrade software? That should be done via scheduled times with a plan in place in case there is an issue and if part of the core OS (postgresql/httpd) should be done via apt-get if at all possible. Sincerely, Joshua D. Drake > > Regards, Dave. > > > ---------------------------(end of broadcast)--------------------------- > TIP 6: explain analyze is your friend -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/
> >> Your requirements as listed above distinctly note a lack of a need > >> for root access. Thus I really don't know what your concern is > >> outside of a feeling that you need "control". > > > > Yes, we do. The community should control things, not > individual companies. > > Are you saying that Command Prompt is not a part of the > community? Or myself for that matter? I don't think that's what he's saying at all. The point I beleive is that your "support staff" is probabliy not reading -www and the other community lists regularly, and therefor not up to date on every part of how things are run. And specificall, I would assume he's referring to the "web community" - as in the ppl working activeliy with the web sites and serviers - which is a lot smaller still. (But hey, we'd welcome additions there) > > None of the other entities providing the project with hosting have > > gone out of their way to deny us the ability to control the > services > > we provide, > > I am not going out of my way. I am not sure why you have such > a problem with this. > > 1. This is the way that Command Prompt, Inc. does hosting for > all people that it hosts. Nobody gets root access. Really? For "server hosting" as well, or do you just do "web hosting"? > 2. You still haven't provided a single reason why you "need" > root access. AFAIK, you need root to restart apache, just as an example. (So it can bind to port 80). Say after a recompile because you needed to tweak a module. Sure, you can set up sudo for each individual command, but that means you have to know everything ahead of time. > > as the people who ultimately have to make sure these things > work we do > > not feel that having to rely on (for example) a level 1 > support tech > > in the middle of the night who has no idea what our server > does or how > > it works is in any way a good thing. > > What level 1 support tech would this be? I don't have any > level 1 support techs. Remember, we are not a hosting company. Whatever people you'd page in the middle of the night if things go down :-) Or is that you all the time? > > This is a dedicated server for the PostgreSQL project, that you > > offered to us knowing full well we were expecting full root > access to. > > Well no. I didn't expect that you would need root access > because I have 50 dedicated machines at the facility none of > which need root access. > > I am not trying to be difficult here but all I see is, "Well > we like to do things as root and since this is only a > community machine you should let us." That's not really so. Maybe there weren't enough details in the original mail. We do beleive there is a *reason* for it. See above for one example. > I on the other hand am trying to bring a certain level of > stability and quality to the infrastructure. That requires a > level of discipline which means we use things like sudo, acls > and group rights. We don't use root. I do beleive we'd be fine without root as long as we could do all the things required - it's not root in the absolute that's needed. This does include recompiling and restarting "line of business" apps like apache. (But not necessariliy things like changing system libs or kernel - I'm fine with some on-site tech dealing with that) > You will need to configure apache... I will make sure you can > do so via included confs. See above - config not enough, recompile/replace needed. For flexibility. (Needed is always a relative matter of course, but it would certainly make things a hell of a lot easier) > You will need to be able to stop/start postgresql. You will > be able to do so via sudo. > > You will need to be able stop/start apache... You will be > able to do so via sudo. > > You will need to be able to add users... I can give you sudo > rights to do that (although that scares me a bit). Well as said above, we can set everything up with sudo. But that cuts down flexibility quite a bit, since every time you need to do something "outside the box", you're stuck. But in general, if you're scared of the people maintaining the other community servers, then perhaps there is a bigger problem... > You want to be able to upgrade software? That should be done > via scheduled times with a plan in place in case there is an > issue and if part of the core OS (postgresql/httpd) should be > done via apt-get if at all possible. For core-os, absolutely. For LOB, we've learned from the stuff we have on the other machines that building them from source is more or less required. Using the pre-packaged ones isn't flexible enough when it comes to which modules are loaded and not. And assuming you test things properly (say on a different port) before you do it, you can do most upgrades with sub-second downtime, so they can be done without scheduling a particular service window. Some things take longer, and need to be scheduled. //Magnus
>>> None of the other entities providing the project with hosting have >>> gone out of their way to deny us the ability to control the >> services >>> we provide, >> I am not going out of my way. I am not sure why you have such >> a problem with this. >> >> 1. This is the way that Command Prompt, Inc. does hosting for >> all people that it hosts. Nobody gets root access. > > Really? For "server hosting" as well, or do you just do "web hosting"? Yes Really. Even for dedicated machines (which is majority of our hosting). Nobody gets root access. > AFAIK, you need root to restart apache, just as an example. (So it can > bind to port 80). Say after a recompile because you needed to tweak a > module. Sure, you can set up sudo for each individual command, but that > means you have to know everything ahead of time. Well I would hope that you would know everything ahead of time. Of course nobody is perfect but if we actually document this machine we shouldn't have any problems. We could even (and I am happy to do this) set up a Trac for the machine so we know what the heck is going on with it. Actually a Trac for all of pgsql-www might be a pretty good idea. >>> it works is in any way a good thing. >> What level 1 support tech would this be? I don't have any >> level 1 support techs. Remember, we are not a hosting company. > > Whatever people you'd page in the middle of the night if things go down > :-) Or is that you all the time? It isn't always me, but it is never a tier 1 and if we do what we are supposed to do :) this will be documented and wouldn't be an issue. >> I am not trying to be difficult here but all I see is, "Well >> we like to do things as root and since this is only a >> community machine you should let us." > > That's not really so. Maybe there weren't enough details in the original > mail. We do beleive there is a *reason* for it. See above for one > example. Well I again refer back to using sudo. >> I on the other hand am trying to bring a certain level of >> stability and quality to the infrastructure. That requires a >> level of discipline which means we use things like sudo, acls >> and group rights. We don't use root. > > I do beleive we'd be fine without root as long as we could do all the > things required - it's not root in the absolute that's needed. This does > include recompiling and restarting "line of business" apps like apache. > (But not necessariliy things like changing system libs or kernel - I'm > fine with some on-site tech dealing with that) Well I would question the need to recompile apache (that is what apxs is for) but I get your point. If it really comes do to you "needing" to recompile apache then we can do what we do for the buildfarm which is proxy to a private apache instance that is completely controlled via userspace. >> You will need to configure apache... I will make sure you can >> do so via included confs. > > See above - config not enough, recompile/replace needed. For > flexibility. (Needed is always a relative matter of course, but it would > certainly make things a hell of a lot easier) A lot of this argument seems to come down to making sure things are available in general. I am one of the most available guys when it comes to the community. I am not hard to find ;). If you need something that isn't there, we can make sure it happens. > Well as said above, we can set everything up with sudo. But that cuts > down flexibility quite a bit, since every time you need to do something > "outside the box", you're stuck. Not stuck, just need to plan :). Again, if we need something outside the box we can make sure that happens. > But in general, if you're scared of the people maintaining the other > community servers, then perhaps there is a bigger problem... It isn't an issue of being scared. It is an issue of liability, security, and insuring a quality of service. The root user is a foot gun. I would prefer you all keep your toes :) > For core-os, absolutely. For LOB, we've learned from the stuff we have > on the other machines that building them from source is more or less > required. Using the pre-packaged ones isn't flexible enough when it > comes to which modules are loaded and not. Well we are getting a little semantic here, and I would actually love to have this discussion on a different thread because my experience is different. (there are exceptions of course) > And assuming you test things properly (say on a different port) before > you do it, you can do most upgrades with sub-second downtime, so they > can be done without scheduling a particular service window. Some things > take longer, and need to be scheduled. It should still be scheduled so people know it is happening "just in case" :) Joshua D. Drake > > > //Magnus -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/
On 14/2/06 22:17, "Joshua D. Drake" <jd@commandprompt.com> wrote: > Are you saying that Command Prompt is not a part of the community? Or > myself for that matter? Of course not. I'm saying that I consider the term community to mean more than 1 entity within our collective group. > 1. This is the way that Command Prompt, Inc. does hosting for all people > that it hosts. Nobody gets root access. But you say below you are not a hosting company?!?! > 2. You still haven't provided a single reason why you "need" root access. Yes I did. Not the best admittedly, but all valid reasons. >> as the people who ultimately have to make sure these things work we do not >> feel that having to rely on (for example) a level 1 support tech in the >> middle of the night who has no idea what our server does or how it works is >> in any way a good thing. > > What level 1 support tech would this be? I don't have any level 1 > support techs. Remember, we are not a hosting company. OK, whoever gets the pleasure of answering the phone at 3AM your time and being told to fix X, Y & Z now because someone has found a way to spam through some bit of sofwatre previously thought to be secure. >> >> This is a dedicated server for the PostgreSQL project, that you offered to >> us knowing full well we were expecting full root access to. > > Well no. I didn't expect that you would need root access because I have > 50 dedicated machines at the facility none of which need root access. http://archives.postgresql.org/pgsql-www/2006-01/msg00119.php > I am not trying to be difficult here but all I see is, "Well we like to > do things as root and since this is only a community machine you should > let us." > > I on the other hand am trying to bring a certain level of stability and > quality to the infrastructure. That requires a level of discipline which > means we use things like sudo, acls and group rights. We don't use root. >> Why is it such a >> bad thing to provide exactly that? > > Dave, come on. You are smarter then that. System Administrator 101 says > you don't use root unless you have too. You don't give root to those who > don't need it. No, we don't. However we do give root to enough trusted individuals with intimate knowledge of the system to ensure it can fulfill al the tasks it is required to at all times. > I refer to #2 above. Nobody has given me one task that you would need to > require root on this machine. OK, example: major disaster on another box- we temporarily need to move another site over which requires an addition PHP option be enabled. That has happened in a situation when we needed to do the job immediately or face being slashdotted on release day. > You will need to configure apache... I will make sure you can do so via > included confs. > > You will need to configure Aspseek... You will be able to do so. > > You will need to be able to stop/start postgresql. You will be able to > do so via sudo. > > You will need to be able stop/start apache... You will be able to do so > via sudo. > > You will need to be able to add users... I can give you sudo rights to > do that (although that scares me a bit). All fine, as long as we know in advance every little thing we might need to do which seems highly unlikely. > You want to be able to upgrade software? That should be done via > scheduled times with a plan in place in case there is an issue and if > part of the core OS (postgresql/httpd) should be done via apt-get if at > all possible. Which is rarely flexible enough based on past experience of various *nixes. I still don't see why this is a problem given that this is a dedicated machine, unless you're scared that we will break it because we don't know what we're doing (human error not being a factor as your people are no more safe from that than we are). Regards, Dave.
> Which is rarely flexible enough based on past experience of various *nixes. > > I still don't see why this is a problem given that this is a dedicated > machine, unless you're scared that we will break it because we don't know > what we're doing (human error not being a factor as your people are no more > safe from that than we are). > > Regards, Dave. I am not scared *you* will break it. I am scared *someone* may break it. The more people that have some level of administrative control the more chance somebody is going to screw it up. I go back to my mention of nobody knowing who installed an alpha version of mailman on pgFoundry. My only real goal here beyond business requirements (and yes they do apply because this machine is on a business network) is to insure we have a very stable and secure environment for search. Unless we decide to move it to a different search platform, this machine should not change -- ever. Production machines don't change. They are static except for the data the provide. There shouldn't be any development on this machine. Sincerely, Joshua D. Drake -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/
>> 1. This is the way that Command Prompt, Inc. does hosting for all people >> that it hosts. Nobody gets root access. > > But you say below you are not a hosting company?!?! I meant we are not a RackSpace or Hub.org. The majority (over 90%) of our revenue does not come from hosting. Hosting is a value add we provide to the community and our customers who require a level of love that other customers don't. Sincerely, Joshua D. Drake -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/
"Joshua D. Drake" <jd@commandprompt.com> writes:
> Unless we decide to move it to a different search platform, this machine
> should not change -- ever. Production machines don't change.
Hmm ... obviously Red Hat is a mirage, because their entire business
model is built on the fact that production machines *do* require
updates. Especially ones that provide services to the net, and
therefore can't be completely hidden behind a firewall.
Somebody's got to be responsible for updating the software on this
server, and being prepared to do so on short notice, and being prepared
to update any part of it from kernel on up.
regards, tom lane
Tom Lane wrote: > "Joshua D. Drake" <jd@commandprompt.com> writes: >> Unless we decide to move it to a different search platform, this machine >> should not change -- ever. Production machines don't change. > > Hmm ... obviously Red Hat is a mirage, because their entire business > model is built on the fact that production machines *do* require > updates. Especially ones that provide services to the net, and > therefore can't be completely hidden behind a firewall. > > Somebody's got to be responsible for updating the software on this > server, and being prepared to do so on short notice, and being prepared > to update any part of it from kernel on up. I was not speaking about updates. I was talking about something such as recompiling apache to add a php module. Of course we have to apply security updates etc.. And we are prepared to do so on short notice, I blindly assumed that CMD would be responsible for a good portion of this as we offered to host the environment in the first place. I wasn't looking for a plug and run(away) solution for the community. Sincerely, Joshua D. Drake > > regards, tom lane -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/
On 14/2/06 23:24, "Joshua D. Drake" <jd@commandprompt.com> wrote: > And we are prepared to do so on short notice, I blindly assumed that CMD > would be responsible for a good portion of this as we offered to host > the environment in the first place. Which is great as far as the core OS goes, and we'd be grateful to CMD for handling it. Our wish for root access is to enable us to deal with non-OS related issues (upgrading a streamlined Apache build for example) and to handle unforseen problems that you or your staff might not be able to deal with immediately, or have enough knowledge of the specifics to deal with without hand-holding. Regards, Dave.
Hi, On Tue, 2006-02-14 at 23:41 +0000, Dave Page wrote: > Which is great as far as the core OS goes, and we'd be grateful to CMD for > handling it. Our wish for root access is to enable us to deal with non-OS > related issues (upgrading a streamlined Apache build for example) and to > handle unforseen problems that you or your staff might not be able to deal > with immediately, or have enough knowledge of the specifics to deal with > without hand-holding. Dave, why don't you use the vendor packages? I believe Joshua will use Fedora Core, and I'd use vendor Apache, etc. For example, today I upgraded RPM's of Ferengi and now one will need to recompile Apache and other stuff as every new version is released -- Why do we lose time with that? As a part of WWW team and CMD, I'd be happy to coordinate the stuff, BTW. Regards, -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: PL/php, plPerlNG - http://www.commandprompt.com/
-----Original Message----- From: Devrim GUNDUZ [mailto:devrim@commandprompt.com] Sent: Wed 2/15/2006 12:50 PM To: Dave Page Cc: Joshua D. Drake; Tom Lane; pgsql-www@postgresql.org Subject: Re: [pgsql-www] Search machine is ready > Dave, why don't you use the vendor packages? I believe Joshua will use > Fedora Core, and I'd use vendor Apache, etc. I was assuming Debian - I could have sworn he mentioned apt at some point. wrt the vendor packages, I do try to use them whereever possible, with a couple of exceptions: - PHP (and occasionally Apache) is often not built the way I want it - either being full of junk, or missing things. I installedPHP5.1 from FreeBSD ports the other day though, and it does seem much more modular there now - dunno if the sameapplies to FC/Debian et al these days. - Vendor changes. I don't like to run packages I know to have been changed by the vendor - for example, it was noted theother day that the Debian PostgreSQL packages contains patches that aren't in PGDG releases. In this case the only OSvendor I would trust with a modified PostgreSQL is Red Hat, for fairly obvious reasons. > As a part of WWW team and CMD, I'd be happy to coordinate the stuff, > BTW. Please feel free :-) Regards,Dave.
Dave Page wrote: > > -----Original Message----- > From: Devrim GUNDUZ [mailto:devrim@commandprompt.com] > Sent: Wed 2/15/2006 12:50 PM > To: Dave Page > Cc: Joshua D. Drake; Tom Lane; pgsql-www@postgresql.org > Subject: Re: [pgsql-www] Search machine is ready > > >> Dave, why don't you use the vendor packages? I believe Joshua will use >> Fedora Core, and I'd use vendor Apache, etc. >> > > I was assuming Debian - I could have sworn he mentioned apt at some point. > Well FYI if we were using Fedora, we would probably use apt anyway ;) That being said... It is Ubuntu Breezy. > - Vendor changes. I don't like to run packages I know to have been changed by the vendor - for example, it was noted theother day that the Debian PostgreSQL packages contains patches that aren't in PGDG releases. In this case the only OSvendor I would trust with a modified PostgreSQL is Red Hat, for fairly obvious reasons. > The search machine currently has MammothPostgreSQL FOSS on it. Which is pure PostgreSQL. > >> As a part of WWW team and CMD, I'd be happy to coordinate the stuff, >> BTW. >> > > Please feel free :-) > > Regards,Dave. > -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: PLphp, PLperl - http://www.commandprompt.com/
> > - Vendor changes. I don't like to run packages I know to > have been changed by the vendor - for example, it was noted > the other day that the Debian PostgreSQL packages contains > patches that aren't in PGDG releases. In this case the only > OS vendor I would trust with a modified PostgreSQL is Red > Hat, for fairly obvious reasons. > > > The search machine currently has MammothPostgreSQL FOSS on > it. Which is pure PostgreSQL. Well, we *certainly* can't trust those darn MammothPG guys to get things right :-) //Magnus
-----Original Message----- From: "Joshua D. Drake"<jd@commandprompt.com> Sent: 15/02/06 17:02:00 To: "Dave Page"<dpage@vale-housing.co.uk> Cc: "Devrim GUNDUZ"<devrim@commandprompt.com>, "Tom Lane"<tgl@sss.pgh.pa.us>, "pgsql-www@postgresql.org"<pgsql-www@postgresql.org> Subject: Re: [pgsql-www] Search machine is ready Dave Page wrote: > > -----Original Message----- > From: Devrim GUNDUZ [mailto:devrim@commandprompt.com] > Sent: Wed 2/15/2006 12:50 PM > To: Dave Page > Cc: Joshua D. Drake; Tom Lane; pgsql-www@postgresql.org > Subject: Re: [pgsql-www] Search machine is ready > > > The search machine currently has > MammothPostgreSQL FOSS on it. Which is > pure PostgreSQL. I'm not worried about you guys - PostgreSQL is what you do - it's A.N. Other distro that might patch their packages withoutfully understanding what they're doing. /D -----Unmodified Original Message----- Dave Page wrote: > > -----Original Message----- > From: Devrim GUNDUZ [mailto:devrim@commandprompt.com] > Sent: Wed 2/15/2006 12:50 PM > To: Dave Page > Cc: Joshua D. Drake; Tom Lane; pgsql-www@postgresql.org > Subject: Re: [pgsql-www] Search machine is ready > > >> Dave, why don't you use the vendor packages? I believe Joshua will use >> Fedora Core, and I'd use vendor Apache, etc. >> > > I was assuming Debian - I could have sworn he mentioned apt at some point. > Well FYI if we were using Fedora, we would probably use apt anyway ;) That being said... It is Ubuntu Breezy. > - Vendor changes. I don't like to run packages I know to have been changed by the vendor - for example, it was noted theother day that the Debian PostgreSQL packages contains patches that aren't in PGDG releases. In this case the only OSvendor I would trust with a modified PostgreSQL is Red Hat, for fairly obvious reasons. > The search machine currently has MammothPostgreSQL FOSS on it. Which is pure PostgreSQL. > >> As a part of WWW team and CMD, I'd be happy to coordinate the stuff, >> BTW. >> > > Please feel free :-) > > Regards,Dave. > -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: PLphp, PLperl - http://www.commandprompt.com/
Am Mittwoch, 15. Februar 2006 17:40 schrieb Dave Page: > - Vendor changes. I don't like to run packages I know to have been changed > by the vendor - for example, it was noted the other day that the Debian > PostgreSQL packages contains patches that aren't in PGDG releases. In this > case the only OS vendor I would trust with a modified PostgreSQL is Red > Hat, for fairly obvious reasons. Clearly, you have never looked into the Debian packages. Otherwise you would have seen that the patches are completely trivial and not related to changing any functionality. -- Peter Eisentraut http://developer.postgresql.org/~petere/
> -----Original Message----- > From: Peter Eisentraut [mailto:peter_e@gmx.net] > Sent: 17 February 2006 14:54 > To: pgsql-www@postgresql.org > Cc: Dave Page > Subject: Re: [pgsql-www] Search machine is ready > > Am Mittwoch, 15. Februar 2006 17:40 schrieb Dave Page: > > - Vendor changes. I don't like to run packages I know to > have been changed > > by the vendor - for example, it was noted the other day > that the Debian > > PostgreSQL packages contains patches that aren't in PGDG > releases. In this > > case the only OS vendor I would trust with a modified > PostgreSQL is Red > > Hat, for fairly obvious reasons. > > Clearly, you have never looked into the Debian packages. > Otherwise you would > have seen that the patches are completely trivial and not > related to changing > any functionality. No I haven't, nor do I have any desire to. Aside from the fact that I was using that only as an example, seeing comments like: http://archives.postgresql.org/pgsql-hackers/2006-02/msg00281.php and http://archives.postgresql.org/pgsql-hackers/2006-02/msg00310.php is enough to put anyone off (or probably should be). As it happens the one I was thinking of was a patch to the Kerberos auth code, which even Tom was unsure of for at least a while from what I recall. If he is unsure about such a 'trivial' patch, then the vast majority of the rest of us probably should be as well. I shall stick to building my own packages for all distros thank you. Regards, Dave.
Am Freitag, 17. Februar 2006 16:12 schrieb Dave Page: > I shall stick to building my own packages for all distros thank you. Please do, but please refrain from spreading FUD and relaying others' FUD. -- Peter Eisentraut http://developer.postgresql.org/~petere/
> -----Original Message----- > From: Peter Eisentraut [mailto:peter_e@gmx.net] > Sent: 17 February 2006 15:55 > To: Dave Page > Cc: pgsql-www@postgresql.org > Subject: Re: [pgsql-www] Search machine is ready > > Am Freitag, 17. Februar 2006 16:12 schrieb Dave Page: > > I shall stick to building my own packages for all distros thank you. > > Please do, but please refrain from spreading FUD and relaying > others' FUD. I don't believe I did - I gave a reason why I prefer my own builds (because I know the OS vendor hasn't patched the code), and referenced one person who has direct experience supporting people with vendors packages, and one who (correctly) said that various distros do indeed distribute patched versions of PostgreSQL. If I quoted or said anything that is innaccurate or untrue then I'll gladly retract it, hang my head in shame and be pelted with virtual rotten tomatoes, but I'm not going to ignore a direct question just in case someone takes the answer as an insult to their OS of choice. You should know me better than that by now :-) Regards, Dave
Peter Eisentraut wrote: > Am Freitag, 17. Februar 2006 16:12 schrieb Dave Page: >> I shall stick to building my own packages for all distros thank you. > > Please do, but please refrain from spreading FUD and relaying others' FUD. Dave Page was not spreading any FUD. Let's not get into a Debian is holier than thou war. He was simply stating his reasons for not wanting to use distribution packages. I believe that is a completely fair and honest item to discuss and has nothing to do with FUD. It does present intelligent conversation about the choices we have as PostgreSQL users. Sincerely, Joshua D. Drake -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: PLphp, PLperl - http://www.commandprompt.com/
Am Freitag, 17. Februar 2006 17:13 schrieb Dave Page: > You should know me better than that by now :-) I do, but I'm afraid others will not, and if they see "PostgreSQL webmaster does not trust anyone but Red Hat" we don't do anyone a service. :) The PostgreSQL packages out there are, with a notable exception starting with G, minimally patched for local circumstances. I would just ask for more informed comments on this that cannot be misread by unsuspecting users. -- Peter Eisentraut http://developer.postgresql.org/~petere/
> -----Original Message----- > From: Peter Eisentraut [mailto:peter_e@gmx.net] > Sent: 17 February 2006 16:55 > To: Dave Page > Cc: pgsql-www@postgresql.org > Subject: Re: [pgsql-www] Search machine is ready > > Am Freitag, 17. Februar 2006 17:13 schrieb Dave Page: > > You should know me better than that by now :-) > > I do, but I'm afraid others will not, and if they see > "PostgreSQL webmaster > does not trust anyone but Red Hat" we don't do anyone a service. :) OK, well, I'll agree that one could be misunderstood, though I still wouldn't call it spreading FUD. s/Red Hat/$VENDOR employing senior PG hacker/g Oh, and FWIW, my distro of choice is Slackware. :-p Regards Dave.
> Oh, and FWIW, my distro of choice is Slackware. :-p O.k. now we have to start a war! Slackware? What are you thinking!!!! ;) Joshua D. Drake > > Regards Dave. > > ---------------------------(end of broadcast)--------------------------- > TIP 1: if posting/reading through Usenet, please send an appropriate > subscribe-nomail command to majordomo@postgresql.org so that your > message can get through to the mailing list cleanly -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/
On Friday 17 February 2006 09:53, Peter Eisentraut wrote:
> Am Mittwoch, 15. Februar 2006 17:40 schrieb Dave Page:
> > - Vendor changes. I don't like to run packages I know to have been
> > changed by the vendor - for example, it was noted the other day that the
> > Debian PostgreSQL packages contains patches that aren't in PGDG releases.
> > In this case the only OS vendor I would trust with a modified PostgreSQL
> > is Red Hat, for fairly obvious reasons.
>
> Clearly, you have never looked into the Debian packages. Otherwise you
> would have seen that the patches are completely trivial and not related to
> changing any functionality.
Clearly somebody hasn't looked into the Debian packages, lest they would be
familiar with the completely debian specific binaries like clusterdb and
pg_wrapper...
--
Robert Treat
Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL
-----Original Message----- From: "Joshua D. Drake"<jd@commandprompt.com> Sent: 17/02/06 17:29:40 To: "Dave Page"<dpage@vale-housing.co.uk> Cc: "Peter Eisentraut"<peter_e@gmx.net>, "pgsql-www@postgresql.org"<pgsql-www@postgresql.org> Subject: Re: [pgsql-www] Search machine is ready > O.k. now we have to start a war! > Slackware? What are you thinking!!!! ;) Security & stability through simplicity (amongst other things of course). I dislike OS's that assume I'm gonna run X, orhave overly complex packaging systems with a tendancy to break, or use complex config files to make things easy for theGUI tools to read & write. With Slack (or Slamd64), it's easy to build a bare bones system that does exactly what's requiredand no more - which also makes it perfect for building single purpose vserver systems. Of course, that's not what everyone wants, but it suits me. /D -----Unmodified Original Message----- > Oh, and FWIW, my distro of choice is Slackware. :-p O.k. now we have to start a war! Slackware? What are you thinking!!!! ;) Joshua D. Drake > > Regards Dave. > > ---------------------------(end of broadcast)--------------------------- > TIP 1: if posting/reading through Usenet, please send an appropriate > subscribe-nomail command to majordomo@postgresql.org so that your > message can get through to the mailing list cleanly -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/
>> O.k. now we have to start a war! >> Slackware? What are you thinking!!!! ;) > > Security & stability through simplicity (amongst other things of course). I dislike OS's that assume I'm gonna run X, orhave overly complex packaging systems with a tendancy to break, or use complex config files to make things easy for theGUI tools to read & write. With Slack (or Slamd64), it's easy to build a bare bones system that does exactly what's requiredand no more - which also makes it perfect for building single purpose vserver systems. > All the reasons I love Ubuntu :). FYI I know of Slack. In fact I was selling 5k webservers with Slack on them when the only choices were Slack and Yggdrasil. Joshua D. Drake -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: plPHP, plPerlNG - http://www.commandprompt.com/
Robert Treat wrote: > Clearly somebody hasn't looked into the Debian packages, lest they > would be familiar with the completely debian specific binaries like > clusterdb and pg_wrapper... Clearly you haven't looked into the PostgreSQL source tree lately or you would have seen clusterdb there. :) I'm not sure what your problem with pg_wrapper is, as it's not in the path and doesn't hurt anyone. Anyway, let's stop this. If someone thinks something is wrong with a package, report a bug there or discuss the specific problems on our mailing lists. -- Peter Eisentraut http://developer.postgresql.org/~petere/
On Friday 17 February 2006 16:44, Peter Eisentraut wrote:
> Robert Treat wrote:
> > Clearly somebody hasn't looked into the Debian packages, lest they
> > would be familiar with the completely debian specific binaries like
> > clusterdb and pg_wrapper...
>
> Clearly you haven't looked into the PostgreSQL source tree lately or you
> would have seen clusterdb there. :)
lol. I was thinking of all the pg_ctlcluster/pg_dropcluster/etc... stuff.
Maybe a better example would be debian releasing a 7.5 version of pg.
(Somehow I can't seem them getting excited about me submitting a bug that
there is no 7.5 pg...)
--
Robert Treat
Build A Brighter Lamp :: Linux Apache {middleware} PostgreSQL
Joshua D. Drake wrote: > > >> 1. This is the way that Command Prompt, Inc. does hosting for all people > >> that it hosts. Nobody gets root access. > > > > But you say below you are not a hosting company?!?! > > I meant we are not a RackSpace or Hub.org. The majority (over 90%) of > our revenue does not come from hosting. Hosting is a value add we > provide to the community and our customers who require a level of love > that other customers don't. Wow, a "level of love". Any pictures? What would that look like exactly? [ponders] :-) -- Bruce Momjian http://candle.pha.pa.us SRA OSS, Inc. http://www.sraoss.com + If your life is a hard drive, Christ can be your backup. +
>> I meant we are not a RackSpace or Hub.org. The majority (over 90%) of >> our revenue does not come from hosting. Hosting is a value add we >> provide to the community and our customers who require a level of love >> that other customers don't. >> > > Wow, a "level of love". Any pictures? What would that look like > exactly? [ponders] :-) > Sorry. Due to possible political interests all photographs have been burned. At least that is what my chief of staff tells me ;) Joshua D. Drake -- The PostgreSQL Company - Command Prompt, Inc. 1.503.667.4564 PostgreSQL Replication, Consulting, Custom Development, 24x7 support Managed Services, Shared and Dedicated Hosting Co-Authors: PLphp, PLperl - http://www.commandprompt.com/
Yay, it's now up and running... Joshua, as I'm unfamilar with the ubuntu init scripts,. Could you please write one to start: searchd -D -R /usr/local/aspseek/etc/search.d/searchd_pg.conf on startup, after postgresql. For the time being,. I will run the update script manually, until I'm satisfied it runs glitch-free. After that, it will need to be run as a cron job hourly. It's '/usr/local/aspseek/etc/update-archives' Kind Regards, John > -----Original Message----- > From: Joshua D. Drake [mailto:jd@commandprompt.com] > Sent: Monday, March 06, 2006 6:49 AM > To: John Hansen > Cc: Dave Page > Subject: Re: [pgsql-www] Search machine is ready > > Good lord one a pain in the ass :) > > I got it compiled and installed ;) it is all in /usr/local/aspseek... > > > >
Ohh, I presume someone will need to update DNS now.... I will leave this side of it running until I see no more traffic. Shouldn't be more than a few days, a week tops. ... John > -----Original Message----- > From: pgsql-www-owner@postgresql.org > [mailto:pgsql-www-owner@postgresql.org] On Behalf Of John Hansen > Sent: Monday, March 13, 2006 1:26 AM > To: Joshua D. Drake > Cc: Dave Page; pgsql-www@postgresql.org > Subject: Re: [pgsql-www] Search machine is ready > > Yay, it's now up and running... > > Joshua, as I'm unfamilar with the ubuntu init scripts,. Could > you please write one to start: > > searchd -D -R /usr/local/aspseek/etc/search.d/searchd_pg.conf > > on startup, after postgresql. > > For the time being,. I will run the update script manually, > until I'm satisfied it runs glitch-free. > After that, it will need to be run as a cron job hourly. > > It's '/usr/local/aspseek/etc/update-archives' > > Kind Regards, > > John > > > -----Original Message----- > > From: Joshua D. Drake [mailto:jd@commandprompt.com] > > Sent: Monday, March 06, 2006 6:49 AM > > To: John Hansen > > Cc: Dave Page > > Subject: Re: [pgsql-www] Search machine is ready > > > > Good lord one a pain in the ass :) > > > > I got it compiled and installed ;) it is all in > /usr/local/aspseek... > > > > > > > > > > ---------------------------(end of > broadcast)--------------------------- > TIP 9: In versions below 8.0, the planner will ignore your desire to > choose an index scan if your joining column's datatypes do not > match > >
How about we set up a temporary "search2" or sometihng and have people pound on it for a couple of days before we switch over? //Magnus > Ohh, > > I presume someone will need to update DNS now.... > I will leave this side of it running until I see no more traffic. > Shouldn't be more than a few days, a week tops. > > ... John > > > -----Original Message----- > > From: pgsql-www-owner@postgresql.org > > [mailto:pgsql-www-owner@postgresql.org] On Behalf Of John Hansen > > Sent: Monday, March 13, 2006 1:26 AM > > To: Joshua D. Drake > > Cc: Dave Page; pgsql-www@postgresql.org > > Subject: Re: [pgsql-www] Search machine is ready > > > > Yay, it's now up and running... > > > > Joshua, as I'm unfamilar with the ubuntu init scripts,. Could you > > please write one to start: > > > > searchd -D -R /usr/local/aspseek/etc/search.d/searchd_pg.conf > > > > on startup, after postgresql. > > > > For the time being,. I will run the update script manually, > until I'm > > satisfied it runs glitch-free. > > After that, it will need to be run as a cron job hourly. > > > > It's '/usr/local/aspseek/etc/update-archives' > > > > Kind Regards, > > > > John > > > > > -----Original Message----- > > > From: Joshua D. Drake [mailto:jd@commandprompt.com] > > > Sent: Monday, March 06, 2006 6:49 AM > > > To: John Hansen > > > Cc: Dave Page > > > Subject: Re: [pgsql-www] Search machine is ready > > > > > > Good lord one a pain in the ass :) > > > > > > I got it compiled and installed ;) it is all in > > /usr/local/aspseek... > > > > > > > > > > > > > > > > ---------------------------(end of > > broadcast)--------------------------- > > TIP 9: In versions below 8.0, the planner will ignore your desire to > > choose an index scan if your joining column's > datatypes do not > > match > > > > > > ---------------------------(end of > broadcast)--------------------------- > TIP 1: if posting/reading through Usenet, please send an appropriate > subscribe-nomail command to majordomo@postgresql.org > so that your > message can get through to the mailing list cleanly >
Works for me... > -----Original Message----- > From: Magnus Hagander [mailto:mha@sollentuna.net] > Sent: Monday, March 13, 2006 1:43 AM > To: John Hansen; Joshua D. Drake > Cc: Dave Page; pgsql-www@postgresql.org > Subject: RE: [pgsql-www] Search machine is ready > > How about we set up a temporary "search2" or sometihng and > have people pound on it for a couple of days before we switch over? > > //Magnus > > > Ohh, > > > > I presume someone will need to update DNS now.... > > I will leave this side of it running until I see no more traffic. > > Shouldn't be more than a few days, a week tops. > > > > ... John > > > > > -----Original Message----- > > > From: pgsql-www-owner@postgresql.org > > > [mailto:pgsql-www-owner@postgresql.org] On Behalf Of John Hansen > > > Sent: Monday, March 13, 2006 1:26 AM > > > To: Joshua D. Drake > > > Cc: Dave Page; pgsql-www@postgresql.org > > > Subject: Re: [pgsql-www] Search machine is ready > > > > > > Yay, it's now up and running... > > > > > > Joshua, as I'm unfamilar with the ubuntu init scripts,. Could you > > > please write one to start: > > > > > > searchd -D -R /usr/local/aspseek/etc/search.d/searchd_pg.conf > > > > > > on startup, after postgresql. > > > > > > For the time being,. I will run the update script manually, > > until I'm > > > satisfied it runs glitch-free. > > > After that, it will need to be run as a cron job hourly. > > > > > > It's '/usr/local/aspseek/etc/update-archives' > > > > > > Kind Regards, > > > > > > John > > > > > > > -----Original Message----- > > > > From: Joshua D. Drake [mailto:jd@commandprompt.com] > > > > Sent: Monday, March 06, 2006 6:49 AM > > > > To: John Hansen > > > > Cc: Dave Page > > > > Subject: Re: [pgsql-www] Search machine is ready > > > > > > > > Good lord one a pain in the ass :) > > > > > > > > I got it compiled and installed ;) it is all in > > > /usr/local/aspseek... > > > > > > > > > > > > > > > > > > > > > > ---------------------------(end of > > > broadcast)--------------------------- > > > TIP 9: In versions below 8.0, the planner will ignore > your desire to > > > choose an index scan if your joining column's > > datatypes do not > > > match > > > > > > > > > > ---------------------------(end of > > broadcast)--------------------------- > > TIP 1: if posting/reading through Usenet, please send an appropriate > > subscribe-nomail command to majordomo@postgresql.org so that > > your > > message can get through to the mailing list cleanly > > > >
Could someone with proper access please add search2 IN A 207.173.203.163 to the postgresql.org DNS. ... John > -----Original Message----- > From: John Hansen > Sent: Monday, March 13, 2006 1:46 AM > To: 'Magnus Hagander'; 'Joshua D. Drake' > Cc: 'Dave Page'; 'pgsql-www@postgresql.org' > Subject: RE: [pgsql-www] Search machine is ready > > Works for me... > > > -----Original Message----- > > From: Magnus Hagander [mailto:mha@sollentuna.net] > > Sent: Monday, March 13, 2006 1:43 AM > > To: John Hansen; Joshua D. Drake > > Cc: Dave Page; pgsql-www@postgresql.org > > Subject: RE: [pgsql-www] Search machine is ready > > > > How about we set up a temporary "search2" or sometihng and > have people > > pound on it for a couple of days before we switch over? > > > > //Magnus > > > > > Ohh, > > > > > > I presume someone will need to update DNS now.... > > > I will leave this side of it running until I see no more traffic. > > > Shouldn't be more than a few days, a week tops. > > > > > > ... John > > > > > > > -----Original Message----- > > > > From: pgsql-www-owner@postgresql.org > > > > [mailto:pgsql-www-owner@postgresql.org] On Behalf Of John Hansen > > > > Sent: Monday, March 13, 2006 1:26 AM > > > > To: Joshua D. Drake > > > > Cc: Dave Page; pgsql-www@postgresql.org > > > > Subject: Re: [pgsql-www] Search machine is ready > > > > > > > > Yay, it's now up and running... > > > > > > > > Joshua, as I'm unfamilar with the ubuntu init scripts,. > Could you > > > > please write one to start: > > > > > > > > searchd -D -R /usr/local/aspseek/etc/search.d/searchd_pg.conf > > > > > > > > on startup, after postgresql. > > > > > > > > For the time being,. I will run the update script manually, > > > until I'm > > > > satisfied it runs glitch-free. > > > > After that, it will need to be run as a cron job hourly. > > > > > > > > It's '/usr/local/aspseek/etc/update-archives' > > > > > > > > Kind Regards, > > > > > > > > John > > > > > > > > > -----Original Message----- > > > > > From: Joshua D. Drake [mailto:jd@commandprompt.com] > > > > > Sent: Monday, March 06, 2006 6:49 AM > > > > > To: John Hansen > > > > > Cc: Dave Page > > > > > Subject: Re: [pgsql-www] Search machine is ready > > > > > > > > > > Good lord one a pain in the ass :) > > > > > > > > > > I got it compiled and installed ;) it is all in > > > > /usr/local/aspseek... > > > > > > > > > > > > > > > > > > > > > > > > > > > > ---------------------------(end of > > > > broadcast)--------------------------- > > > > TIP 9: In versions below 8.0, the planner will ignore > > your desire to > > > > choose an index scan if your joining column's > > > datatypes do not > > > > match > > > > > > > > > > > > > > ---------------------------(end of > > > broadcast)--------------------------- > > > TIP 1: if posting/reading through Usenet, please send an > appropriate > > > subscribe-nomail command to > majordomo@postgresql.org so that > > > your > > > message can get through to the mailing list cleanly > > > > > > >
On Sunday 12 March 2006 06:42, Magnus Hagander wrote: > How about we set up a temporary "search2" or sometihng and have people > pound on it for a couple of days before we switch over? That is probably a good idea. Lets make sure the machine can handle it. (If it can't I have a bigger one we can use ;)). I will add the searchd init later today. Joshua D. Drake > > //Magnus > > > Ohh, > > > > I presume someone will need to update DNS now.... > > I will leave this side of it running until I see no more traffic. > > Shouldn't be more than a few days, a week tops. > > > > ... John > > > > > -----Original Message----- > > > From: pgsql-www-owner@postgresql.org > > > [mailto:pgsql-www-owner@postgresql.org] On Behalf Of John Hansen > > > Sent: Monday, March 13, 2006 1:26 AM > > > To: Joshua D. Drake > > > Cc: Dave Page; pgsql-www@postgresql.org > > > Subject: Re: [pgsql-www] Search machine is ready > > > > > > Yay, it's now up and running... > > > > > > Joshua, as I'm unfamilar with the ubuntu init scripts,. Could you > > > please write one to start: > > > > > > searchd -D -R /usr/local/aspseek/etc/search.d/searchd_pg.conf > > > > > > on startup, after postgresql. > > > > > > For the time being,. I will run the update script manually, > > > > until I'm > > > > > satisfied it runs glitch-free. > > > After that, it will need to be run as a cron job hourly. > > > > > > It's '/usr/local/aspseek/etc/update-archives' > > > > > > Kind Regards, > > > > > > John > > > > > > > -----Original Message----- > > > > From: Joshua D. Drake [mailto:jd@commandprompt.com] > > > > Sent: Monday, March 06, 2006 6:49 AM > > > > To: John Hansen > > > > Cc: Dave Page > > > > Subject: Re: [pgsql-www] Search machine is ready > > > > > > > > Good lord one a pain in the ass :) > > > > > > > > I got it compiled and installed ;) it is all in > > > > > > /usr/local/aspseek... > > > > > > > > > > > > > > > > > > ---------------------------(end of > > > broadcast)--------------------------- > > > TIP 9: In versions below 8.0, the planner will ignore your desire to > > > choose an index scan if your joining column's > > > > datatypes do not > > > > > match > > > > ---------------------------(end of > > broadcast)--------------------------- > > TIP 1: if posting/reading through Usenet, please send an appropriate > > subscribe-nomail command to majordomo@postgresql.org > > so that your > > message can get through to the mailing list cleanly > > ---------------------------(end of broadcast)--------------------------- > TIP 4: Have you searched our list archives? > > http://archives.postgresql.org
Added ... On Mon, 13 Mar 2006, John Hansen wrote: > Could someone with proper access please add > > search2 IN A 207.173.203.163 > > to the postgresql.org DNS. > > ... John > >> -----Original Message----- >> From: John Hansen >> Sent: Monday, March 13, 2006 1:46 AM >> To: 'Magnus Hagander'; 'Joshua D. Drake' >> Cc: 'Dave Page'; 'pgsql-www@postgresql.org' >> Subject: RE: [pgsql-www] Search machine is ready >> >> Works for me... >> >>> -----Original Message----- >>> From: Magnus Hagander [mailto:mha@sollentuna.net] >>> Sent: Monday, March 13, 2006 1:43 AM >>> To: John Hansen; Joshua D. Drake >>> Cc: Dave Page; pgsql-www@postgresql.org >>> Subject: RE: [pgsql-www] Search machine is ready >>> >>> How about we set up a temporary "search2" or sometihng and >> have people >>> pound on it for a couple of days before we switch over? >>> >>> //Magnus >>> >>>> Ohh, >>>> >>>> I presume someone will need to update DNS now.... >>>> I will leave this side of it running until I see no more traffic. >>>> Shouldn't be more than a few days, a week tops. >>>> >>>> ... John >>>> >>>>> -----Original Message----- >>>>> From: pgsql-www-owner@postgresql.org >>>>> [mailto:pgsql-www-owner@postgresql.org] On Behalf Of John Hansen >>>>> Sent: Monday, March 13, 2006 1:26 AM >>>>> To: Joshua D. Drake >>>>> Cc: Dave Page; pgsql-www@postgresql.org >>>>> Subject: Re: [pgsql-www] Search machine is ready >>>>> >>>>> Yay, it's now up and running... >>>>> >>>>> Joshua, as I'm unfamilar with the ubuntu init scripts,. >> Could you >>>>> please write one to start: >>>>> >>>>> searchd -D -R /usr/local/aspseek/etc/search.d/searchd_pg.conf >>>>> >>>>> on startup, after postgresql. >>>>> >>>>> For the time being,. I will run the update script manually, >>>> until I'm >>>>> satisfied it runs glitch-free. >>>>> After that, it will need to be run as a cron job hourly. >>>>> >>>>> It's '/usr/local/aspseek/etc/update-archives' >>>>> >>>>> Kind Regards, >>>>> >>>>> John >>>>> >>>>>> -----Original Message----- >>>>>> From: Joshua D. Drake [mailto:jd@commandprompt.com] >>>>>> Sent: Monday, March 06, 2006 6:49 AM >>>>>> To: John Hansen >>>>>> Cc: Dave Page >>>>>> Subject: Re: [pgsql-www] Search machine is ready >>>>>> >>>>>> Good lord one a pain in the ass :) >>>>>> >>>>>> I got it compiled and installed ;) it is all in >>>>> /usr/local/aspseek... >>>>>> >>>>>> >>>>>> >>>>>> >>>>> >>>>> ---------------------------(end of >>>>> broadcast)--------------------------- >>>>> TIP 9: In versions below 8.0, the planner will ignore >>> your desire to >>>>> choose an index scan if your joining column's >>>> datatypes do not >>>>> match >>>>> >>>>> >>>> >>>> ---------------------------(end of >>>> broadcast)--------------------------- >>>> TIP 1: if posting/reading through Usenet, please send an >> appropriate >>>> subscribe-nomail command to >> majordomo@postgresql.org so that >>>> your >>>> message can get through to the mailing list cleanly >>>> >>> >>> > ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
Joshua D. Drake Wrote: > That is probably a good idea. Lets make sure the machine can > handle it. Searches are not going to be a problem on this box :) It's the indexing that is pretty resource intensive....
Okay folks... http://search2.postgresql.org/ is now ready for testing... Please report any issues you may find,.. Kind Regards, John Hansen
> -----Original Message----- > From: pgsql-www-owner@postgresql.org > [mailto:pgsql-www-owner@postgresql.org] On Behalf Of John Hansen > Sent: 13 March 2006 04:40 > To: pgsql-www@postgresql.org > Subject: Re: [pgsql-www] Search machine is ready > > Okay folks... > > > http://search2.postgresql.org/ is now ready for testing... > > Please report any issues you may find,.. "Firefox can't find the server at search2.postgresql.org" :-( /D
> > Okay folks... > > > > > > http://search2.postgresql.org/ is now ready for testing... > > > > Please report any issues you may find,.. > > "Firefox can't find the server at search2.postgresql.org" > > :-( That would be that thing called DNS caching with timeout... Be patient! //MAgnus
> -----Original Message----- > From: Magnus Hagander [mailto:mha@sollentuna.net] > Sent: 13 March 2006 08:45 > To: Dave Page; John Hansen; pgsql-www@postgresql.org > Subject: RE: [pgsql-www] Search machine is ready > > > > Okay folks... > > > > > > > > > http://search2.postgresql.org/ is now ready for testing... > > > > > > Please report any issues you may find,.. > > > > "Firefox can't find the server at search2.postgresql.org" > > > > :-( > > That would be that thing called DNS caching with timeout... > Be patient! I wasn't aware that non-existence of records was cached. /D
> > > > Okay folks... > > > > > > > > > > > > http://search2.postgresql.org/ is now ready for testing... > > > > > > > > Please report any issues you may find,.. > > > > > > "Firefox can't find the server at search2.postgresql.org" > > > > > > :-( > > > > That would be that thing called DNS caching with timeout... > > Be patient! > > I wasn't aware that non-existence of records was cached. Not on the client (actually they are, but generally just for a minute or five), but I bet the whole zone is cached at your resolver... If you point to a hub.org DNS server it's there. But actually - it's not on the lerctr.org servers. Perhaps they don't respond to NOTIFYs or something? //Magnus
Hmmm,.... Odd The expire timeout for the SOA record is 5 hrs. It was added way longer ago than that. My own DNS has picked it up, but I know my ISP's DNS hasn't,. (and hence I can't test it myself yet, as they do transparent proxying) Would most ISP's be conservative and specify a minimum TTL for ALL records, regardless of what they have been configuredto? ... John > -----Original Message----- > From: Magnus Hagander [mailto:mha@sollentuna.net] > Sent: Monday, March 13, 2006 7:45 PM > To: Dave Page; John Hansen; pgsql-www@postgresql.org > Subject: RE: [pgsql-www] Search machine is ready > > > > Okay folks... > > > > > > > > > http://search2.postgresql.org/ is now ready for testing... > > > > > > Please report any issues you may find,.. > > > > "Firefox can't find the server at search2.postgresql.org" > > > > :-( > > That would be that thing called DNS caching with timeout... > Be patient! > > //MAgnus > >
> But actually - it's not on the lerctr.org servers. Perhaps > they don't respond to NOTIFYs or something? Well, that would explain why my ISP hasn't picked it up yet, since lerctr.org are primary for the domain.
> -----Original Message----- > From: Magnus Hagander [mailto:mha@sollentuna.net] > Sent: 13 March 2006 08:58 > To: Dave Page; John Hansen; pgsql-www@postgresql.org > Cc: Larry Rosenman > Subject: RE: [pgsql-www] Search machine is ready > > But actually - it's not on the lerctr.org servers. Perhaps they don't > respond to NOTIFYs or something? That might be the problem - I've seen similar issues when Marc has added A's for me in the past, but only with postgresql.org. Regards, Dave.
Okay, Can’t wait for this DNS business to clean itself up, so, I've made search2.postgresql.org the default site for the server,at least then you can connect on the IP as http://207.173.203.163/archives.search That should be enough for testing purposes... ... John
> -----Original Message----- > From: pgsql-www-owner@postgresql.org > [mailto:pgsql-www-owner@postgresql.org] On Behalf Of John Hansen > Sent: 13 March 2006 12:12 > To: pgsql-www@postgresql.org > Subject: Re: [pgsql-www] Search machine is ready > > Okay, > > Can't wait for this DNS business to clean itself up, so, I've > made search2.postgresql.org the default site for the server, > at least then you can connect on the IP as > > http://207.173.203.163/archives.search > > That should be enough for testing purposes... Yay, now that's speedy. It does seem to be left justified for some reason though. /D
Dave Page Wrote: > Yay, now that's speedy. Yea,. Faster connection makes a difference, tho it is a fraction slower than my box, but we're talking milliseconds on theinitial query, microseconds on subsequent queries.... > It does seem to be left justified for some reason though. What do you mean by 'left justified'? Is it any different than http://search.postgresql.org ? I seem to get the same look from both? ... John
> -----Original Message----- > From: John Hansen [mailto:john@geeknet.com.au] > Sent: 13 March 2006 12:40 > To: Dave Page; pgsql-www@postgresql.org > Subject: RE: [pgsql-www] Search machine is ready > > Dave Page Wrote: > > > Yay, now that's speedy. > > Yea,. Faster connection makes a difference, tho it is a > fraction slower than my box, but we're talking milliseconds > on the initial query, microseconds on subsequent queries.... > > > It does seem to be left justified for some reason though. > > What do you mean by 'left justified'? The page content is not in the middle, like www.postgresql.org. > Is it any different than http://search.postgresql.org ? No, that seems to be borked as well :-( /D
Dave Page Wrote: > > What do you mean by 'left justified'? > > The page content is not in the middle, like www.postgresql.org. > > > Is it any different than http://search.postgresql.org ? > > No, that seems to be borked as well :-( Well, there you go,... The templates are in /usr/local/aspseek/htdocs if you wanna have a fiddle... ... John
> -----Original Message----- > From: John Hansen [mailto:john@geeknet.com.au] > Sent: 13 March 2006 12:49 > To: Dave Page; pgsql-www@postgresql.org > Subject: RE: [pgsql-www] Search machine is ready > > Dave Page Wrote: > > > What do you mean by 'left justified'? > > > > The page content is not in the middle, like www.postgresql.org. > > > > > Is it any different than http://search.postgresql.org ? > > > > No, that seems to be borked as well :-( > > Well, there you go,... > The templates are in /usr/local/aspseek/htdocs if you wanna > have a fiddle... 'k /D
Dave Page wrote: >> -----Original Message----- >> From: Magnus Hagander [mailto:mha@sollentuna.net] >> Sent: 13 March 2006 08:58 >> To: Dave Page; John Hansen; pgsql-www@postgresql.org >> Cc: Larry Rosenman >> Subject: RE: [pgsql-www] Search machine is ready >> >> But actually - it's not on the lerctr.org servers. Perhaps they don't >> respond to NOTIFYs or something? > > That might be the problem - I've seen similar issues when Marc has > added A's for me in the past, but only with postgresql.org. > > Regards, Dave. here's why: -- Larry Rosenman http://www.lerctr.org/~ler Phone: +1 512-248-2683 E-Mail: ler@lerctr.org US Mail: 430 Valona Loop, Round Rock, TX 78681-3893
Magnus Hagander wrote: >>>>> Okay folks... >>>>> >>>>> >>>>> http://search2.postgresql.org/ is now ready for testing... >>>>> >>>>> Please report any issues you may find,.. >>>> >>>> "Firefox can't find the server at search2.postgresql.org" >>>> >>>> :-( >>> >>> That would be that thing called DNS caching with timeout... >>> Be patient! >> >> I wasn't aware that non-existence of records was cached. > > Not on the client (actually they are, but generally just for a minute > or five), but I bet the whole zone is cached at your resolver... > > If you point to a hub.org DNS server it's there. > > But actually - it's not on the lerctr.org servers. Perhaps they don't > respond to NOTIFYs or something? > > //Magnus Am I pointing at the wrong master: Mar 12 22:59:41 lerami named[1352]: zone postgresql.org/IN: refused notify from non-master: 200.46.208.251#2330 Mar 12 22:59:50 lerami named[1352]: client 200.46.204.254#4289: received notify for zone 'postgresql.org' //PostgreSQL Global Development Group/Marc Fornier (scrappy@hub.org) zone "postgresql.org" { type slave; file "cache/postgresql.org"; masters { 200.46.204.2; }; }; zone "mirrors.postgresql.org" { type slave; file "cache/mirrors.postgresql.org"; masters { 62.65.68.81; }; }; -- Larry Rosenman http://www.lerctr.org/~ler Phone: +1 512-248-2683 E-Mail: ler@lerctr.org US Mail: 430 Valona Loop, Round Rock, TX 78681-3893
On Mon, 13 Mar 2006, Larry Rosenman wrote: > Magnus Hagander wrote: >>>>>> Okay folks... >>>>>> >>>>>> >>>>>> http://search2.postgresql.org/ is now ready for testing... >>>>>> >>>>>> Please report any issues you may find,.. >>>>> >>>>> "Firefox can't find the server at search2.postgresql.org" >>>>> >>>>> :-( >>>> >>>> That would be that thing called DNS caching with timeout... >>>> Be patient! >>> >>> I wasn't aware that non-existence of records was cached. >> >> Not on the client (actually they are, but generally just for a minute >> or five), but I bet the whole zone is cached at your resolver... >> >> If you point to a hub.org DNS server it's there. >> >> But actually - it's not on the lerctr.org servers. Perhaps they don't >> respond to NOTIFYs or something? >> >> //Magnus > Am I pointing at the wrong master: > Mar 12 22:59:41 lerami named[1352]: zone postgresql.org/IN: refused notify > from non-master: 200.46.208.251#2330 > Mar 12 22:59:50 lerami named[1352]: client 200.46.204.254#4289: received > notify for zone 'postgresql.org' > > //PostgreSQL Global Development Group/Marc Fornier (scrappy@hub.org) > zone "postgresql.org" { > type slave; > file "cache/postgresql.org"; > masters { > 200.46.204.2; > }; > }; Ya, master has been 200.46.204.13 for awhile now ... ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email: scrappy@hub.org Yahoo!: yscrappy ICQ: 7615664
Marc G. Fournier wrote: > On Mon, 13 Mar 2006, Larry Rosenman wrote: > >> Magnus Hagander wrote: >>>>>>> Okay folks... >>>>>>> >>>>>>> >>>>>>> http://search2.postgresql.org/ is now ready for testing... >>>>>>> >>>>>>> Please report any issues you may find,.. >>>>>> >>>>>> "Firefox can't find the server at search2.postgresql.org" >>>>>> >>>>>> :-( >>>>> >>>>> That would be that thing called DNS caching with timeout... >>>>> Be patient! >>>> >>>> I wasn't aware that non-existence of records was cached. >>> >>> Not on the client (actually they are, but generally just for a >>> minute or five), but I bet the whole zone is cached at your >>> resolver... >>> >>> If you point to a hub.org DNS server it's there. >>> >>> But actually - it's not on the lerctr.org servers. Perhaps they >>> don't respond to NOTIFYs or something? >>> >>> //Magnus >> Am I pointing at the wrong master: >> Mar 12 22:59:41 lerami named[1352]: zone postgresql.org/IN: refused >> notify from non-master: 200.46.208.251#2330 >> Mar 12 22:59:50 lerami named[1352]: client 200.46.204.254#4289: >> received notify for zone 'postgresql.org' >> >> //PostgreSQL Global Development Group/Marc Fornier (scrappy@hub.org) >> zone "postgresql.org" { >> type slave; >> file "cache/postgresql.org"; >> masters { >> 200.46.204.2; >> }; >> }; > > Ya, master has been 200.46.204.13 for awhile now ... > > > ---- > Marc G. Fournier Hub.Org Networking Services > (http://www.hub.org) Email: scrappy@hub.org Yahoo!: > yscrappy ICQ: 7615664 Updated, and refreshed. Ya gotta tell me these things..... LER -- Larry Rosenman http://www.lerctr.org/~ler Phone: +1 512-248-2683 E-Mail: ler@lerctr.org US Mail: 430 Valona Loop, Round Rock, TX 78681-3893
Guys, The indexing part of things seems to be running smoothly. I've done a bit of testing myself, and searches seems fast enough. Nothing more for me to do, so I'm quite happy to switch over at this stage. ... John