Thread: pgfoundry is down
pgfoundry is down Kris Jurka
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 - --On Friday, November 09, 2007 14:22:24 -0500 Kris Jurka <books@ejurka.com> wrote: > > pgfoundry is down Yup, I'm moving it to IO so taht I can reformat the server it is on now to free up 75G of disk space so that nagios doesn't keep complaining about low disk space ... should be back up in a few minutes, just finishing a final rsync ... - ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email . scrappy@hub.org MSN . scrappy@hub.org Yahoo . yscrappy Skype: hub.org ICQ . 7615664 -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4 (FreeBSD) iD8DBQFHNLS34QvfyHIvDvMRAlRjAJ4vGdak4R4C50Kqlsenlf5P4DWodQCgyqQy 4kz0xRbLwVglCh0rxjS+RxU= =+Vt0 -----END PGP SIGNATURE-----
On Fri, Nov 09, 2007 at 03:27:51PM -0400, Marc G. Fournier wrote: > - --On Friday, November 09, 2007 14:22:24 -0500 Kris Jurka <books@ejurka.com> > wrote: > > > > > pgfoundry is down > > Yup, I'm moving it to IO so taht I can reformat the server it is on now to free > up 75G of disk space so that nagios doesn't keep complaining about low disk > space ... I don't mean to be cranky, but it'd have been nice to accomplish all the work needed for pgfoundry _before_ shutting down gborg. From watching much of the recent events from the sidelines, one might be excused in forming the impression that many things are happening as emergencies. We seem to have an awful lot of emergencies lately. Is something going on? A -- Andrew Sullivan Old sigs will return after re-constitution of magic smoke
Marc G. Fournier wrote: > > > --On Friday, November 09, 2007 14:22:24 -0500 Kris Jurka <books@ejurka.com> > wrote: > >> pgfoundry is down > > Yup, I'm moving it to IO so taht I can reformat the server it is on now to free > up 75G of disk space so that nagios doesn't keep complaining about low disk > space ... > > should be back up in a few minutes, just finishing a final rsync ... Uh, I thought you said no pg servers would be affected by that? Did I misunderstand you, or did something happen along the way? //Magnus
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 And back up again ... I'll be taking it down again, hopefully on Saturday / Sunday to move it back to neptune so that I can do the same to IO ... - --On Friday, November 09, 2007 15:27:51 -0400 "Marc G. Fournier" <scrappy@hub.org> wrote: > -----BEGIN PGP SIGNED MESSAGE----- > Hash: SHA1 > > > > - --On Friday, November 09, 2007 14:22:24 -0500 Kris Jurka <books@ejurka.com> > wrote: > >> >> pgfoundry is down > > Yup, I'm moving it to IO so taht I can reformat the server it is on now to > free up 75G of disk space so that nagios doesn't keep complaining about low > disk space ... > > should be back up in a few minutes, just finishing a final rsync ... > > - ---- > Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) > Email . scrappy@hub.org MSN . scrappy@hub.org > Yahoo . yscrappy Skype: hub.org ICQ . 7615664 > -----BEGIN PGP SIGNATURE----- > Version: GnuPG v2.0.4 (FreeBSD) > > iD8DBQFHNLS34QvfyHIvDvMRAlRjAJ4vGdak4R4C50Kqlsenlf5P4DWodQCgyqQy > 4kz0xRbLwVglCh0rxjS+RxU= > =+Vt0 > -----END PGP SIGNATURE----- > > > ---------------------------(end of broadcast)--------------------------- > TIP 6: explain analyze is your friend - ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email . scrappy@hub.org MSN . scrappy@hub.org Yahoo . yscrappy Skype: hub.org ICQ . 7615664 -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4 (FreeBSD) iD8DBQFHNLX34QvfyHIvDvMRAokGAKC4/NBDsxoLr2mqReIe6DO4QZqgngCgioJ0 rBGhOVYioAXL/lBWlOxSAto= =0BI2 -----END PGP SIGNATURE-----
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 - --On Friday, November 09, 2007 20:32:19 +0100 Magnus Hagander <magnus@hagander.net> wrote: > Marc G. Fournier wrote: >> >> >> --On Friday, November 09, 2007 14:22:24 -0500 Kris Jurka <books@ejurka.com> >> wrote: >> >>> pgfoundry is down >> >> Yup, I'm moving it to IO so taht I can reformat the server it is on now to >> free up 75G of disk space so that nagios doesn't keep complaining about low >> disk space ... >> >> should be back up in a few minutes, just finishing a final rsync ... > > Uh, I thought you said no pg servers would be affected by that? Did I > misunderstand you, or did something happen along the way? I had to move it off of the machine I was reformatting ... else we'd lose everything, no? :) - ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email . scrappy@hub.org MSN . scrappy@hub.org Yahoo . yscrappy Skype: hub.org ICQ . 7615664 -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4 (FreeBSD) iD8DBQFHNLkP4QvfyHIvDvMRAvo0AKDrv9NZsLEyOwYCA2mphwXJDqthTgCcCcBE WKQpAiTlmZjj6gyD3pHGT8M= =saUh -----END PGP SIGNATURE-----
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 - --On Friday, November 09, 2007 14:31:16 -0500 Andrew Sullivan <ajs@crankycanuck.ca> wrote: > On Fri, Nov 09, 2007 at 03:27:51PM -0400, Marc G. Fournier wrote: >> - --On Friday, November 09, 2007 14:22:24 -0500 Kris Jurka >> <books@ejurka.com> wrote: >> >> > >> > pgfoundry is down >> >> Yup, I'm moving it to IO so taht I can reformat the server it is on now to >> free up 75G of disk space so that nagios doesn't keep complaining about low >> disk space ... > > I don't mean to be cranky, but it'd have been nice to accomplish all the > work needed for pgfoundry _before_ shutting down gborg. PgFoundry was down for ~10 minutes to move it from one server to the other server oso that the server it was on could be reformatted ... gborg was on that same server, so if gborg was 'still up', it too would have had to be moved ... - ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email . scrappy@hub.org MSN . scrappy@hub.org Yahoo . yscrappy Skype: hub.org ICQ . 7615664 -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4 (FreeBSD) iD8DBQFHNLlj4QvfyHIvDvMRAu2+AKDeUl8ZE7yJKTnJ6dWcTj7WVOmItgCeLc2z OxxpjjosSfioPkH7dj76yzg= =G6SN -----END PGP SIGNATURE-----
On Fri, Nov 09, 2007 at 03:47:47PM -0400, Marc G. Fournier wrote: > > PgFoundry was down for ~10 minutes to move it from one server to the other > server oso that the server it was on could be reformatted ... gborg was on that > same server, so if gborg was 'still up', it too would have had to be moved ... So am I just missing the announcements of these planned outages? It wouldn't surprise me if I were. But if not, then I think we need to come up with some sort of plan of how such things are to be done in the future. There are three possibilities. One is that these are all announced correctly, and I'm just missing them somehow, in which case ignore me. Another is that the recent activities were _planned_ outages, in which case I suggest that some more effort needs to go into informing the community about deadlines, schedules, and service outages. The third is that things are happening in response to emergencies, in which case I want to know what we can do to stop having emergencies. A -- Andrew Sullivan Old sigs will return after re-constitution of blue smoke
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 I put out an announcement several days ago that I would be doing work on neptune that would cause some disruptions to clients on Saturday (ie. extended downtime) ... Magnus confirmed with me that this wouldn't affect any of the Pg VPSs, but I wasn't clear in that although there would be no extended downtime for the Pg VPSs, that I still had to move the affected Pg VPSs *off* of the server being worked on ... my fault, I should have been more clear on that point ... - --On Friday, November 09, 2007 14:56:29 -0500 Andrew Sullivan <ajs@crankycanuck.ca> wrote: > On Fri, Nov 09, 2007 at 03:47:47PM -0400, Marc G. Fournier wrote: >> >> PgFoundry was down for ~10 minutes to move it from one server to the other >> server oso that the server it was on could be reformatted ... gborg was on >> that same server, so if gborg was 'still up', it too would have had to be >> moved ... > > So am I just missing the announcements of these planned outages? It > wouldn't surprise me if I were. But if not, then I think we need to come up > with some sort of plan of how such things are to be done in the future. > > There are three possibilities. One is that these are all announced > correctly, and I'm just missing them somehow, in which case ignore me. > Another is that the recent activities were _planned_ outages, in which case > I suggest that some more effort needs to go into informing the community > about deadlines, schedules, and service outages. The third is that things > are happening in response to emergencies, in which case I want to know what > we can do to stop having emergencies. > > A > > -- > Andrew Sullivan > Old sigs will return after re-constitution of blue smoke > > ---------------------------(end of broadcast)--------------------------- > TIP 4: Have you searched our list archives? > > http://archives.postgresql.org - ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email . scrappy@hub.org MSN . scrappy@hub.org Yahoo . yscrappy Skype: hub.org ICQ . 7615664 -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4 (FreeBSD) iD8DBQFHNL3P4QvfyHIvDvMRAslnAKC+nNe1/qo0suWkne/xcVl/TJHtMgCdGgc/ dVpXEvvls23k+HzseBPhb9k= =SXVi -----END PGP SIGNATURE-----
Marc, > I put out an announcement several days ago that I would be doing work on > neptune that would cause some disruptions to clients on Saturday (ie. > extended downtime) ... Magnus confirmed with me that this wouldn't affect > any of the Pg VPSs, but I wasn't clear in that although there would be no > extended downtime for the Pg VPSs, that I still had to move the affected Pg > VPSs *off* of the server being worked on ... my fault, I should have been > more clear on that point ... It would be nice to have a "planned outages" page somewhere which people could check. It would save us a bunch of answering e-mail, if nothing else. And the outage announcements could get carred in PWN if Fetter is amenable. Question is, where should we host that page? -- Josh Berkus PostgreSQL @ Sun San Francisco
On Tue, Nov 13, 2007 at 09:39:44AM -0800, Josh Berkus wrote: > Marc, > > > I put out an announcement several days ago that I would be doing work on > > neptune that would cause some disruptions to clients on Saturday (ie. > > extended downtime) ... Magnus confirmed with me that this wouldn't affect > > any of the Pg VPSs, but I wasn't clear in that although there would be no > > extended downtime for the Pg VPSs, that I still had to move the affected Pg > > VPSs *off* of the server being worked on ... my fault, I should have been > > more clear on that point ... > > It would be nice to have a "planned outages" page somewhere which people could > check. It would save us a bunch of answering e-mail, if nothing else. And > the outage announcements could get carred in PWN if Fetter is amenable. > > Question is, where should we host that page? I'da say there are two options - either on the main website, which is mirrored globally, or on a completely independent system (not just it's own vm, not even in the same datacenter as our other servers). If we're gonig to do it, let's do it for real :-) //Magnus
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Wed, 14 Nov 2007 09:54:14 +0100
Magnus Hagander <magnus@hagander.net> wrote:
hould we host that page?
>
> I'da say there are two options - either on the main website, which is
> mirrored globally, or on a completely independent system (not just
> it's own vm, not even in the same datacenter as our other servers).
>
> If we're gonig to do it, let's do it for real :-)
What about pmt? Or even pgweb (since that wiki is open).
Joshua D. Drake
- --
=== The PostgreSQL Company: Command Prompt, Inc. ===
Sales/Support: +1.503.667.4564 24x7/Emergency: +1.800.492.2240
PostgreSQL solutions since 1997 http://www.commandprompt.com/ UNIQUE NOT NULL
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate
PostgreSQL Replication: http://www.commandprompt.com/products/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQFHOzKoATb/zqfZUUQRArUQAKCRBciETEXp63HISpCZ9ORI88cqMwCfYsfr
uYcvm197wszGjPrrrSYsD9k=
=hun+
-----END PGP SIGNATURE-----
On Wed, Nov 14, 2007 at 09:38:48AM -0800, Joshua D. Drake wrote: > > On Wed, 14 Nov 2007 09:54:14 +0100 > Magnus Hagander <magnus@hagander.net> wrote: > hould we host that page? > > > > I'da say there are two options - either on the main website, which is > > mirrored globally, or on a completely independent system (not just > > it's own vm, not even in the same datacenter as our other servers). > > > > If we're gonig to do it, let's do it for real :-) > > What about pmt? Or even pgweb (since that wiki is open). pgweb could work. But it's on the same datacenter as some other things, so it would be a bad place to say "archives is down" or "search is down" or so. //Magnus
-----BEGIN PGP SIGNED MESSAGE----- Hash: RIPEMD160 > Question is, where should we host that page? > I'da say there are two options - either on the main website, which is > mirrored globally, or on a completely independent system (not just it's own > vm, not even in the same datacenter as our other servers). +1 for independence. Sound to me like a wiki would be perfect for this as well. My company would be happy to host a wiki (or we could port the existing one over). - -- Greg Sabino Mullane greg@turnstep.com End Point Corporation PGP Key: 0x14964AC8 200711141241 http://biglumber.com/x/web?pk=2529DF6AB8F79407E94445B4BC9B906714964AC8 -----BEGIN PGP SIGNATURE----- iD8DBQFHOzNXvJuQZxSWSsgRA5shAKD7h1NVkAhjvFW/t1amjsbi6z8QowCgyqNe J/Qsp0vhWbaMuhg2kcdi57Q= =77t2 -----END PGP SIGNATURE-----
On Wed, Nov 14, 2007 at 05:42:44PM -0000, Greg Sabino Mullane wrote: > > > > Question is, where should we host that page? > > > I'da say there are two options - either on the main website, which is > > mirrored globally, or on a completely independent system (not just it's own > > vm, not even in the same datacenter as our other servers). > > +1 for independence. Sound to me like a wiki would be perfect for this > as well. My company would be happy to host a wiki (or we could port the > existing one over) If you port the exiting wiki over, you lose independence, no? Also, I think a wiki is a major overkill. All we need is a static webpage that the infrastructure folks can edit, no? Sure, that can be implemented by a wiki, but it just seems way way more complicated than needed. //Magnus
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Wed, 14 Nov 2007 19:23:41 +0100
Magnus Hagander <magnus@hagander.net> wrote:
> On Wed, Nov 14, 2007 at 05:42:44PM -0000, Greg Sabino Mullane wrote:
> >
> >
> > > Question is, where should we host that page?
> >
> > > I'da say there are two options - either on the main website,
> > > which is mirrored globally, or on a completely independent system
> > > (not just it's own vm, not even in the same datacenter as our
> > > other servers).
> >
> > +1 for independence. Sound to me like a wiki would be perfect for
> > this as well. My company would be happy to host a wiki (or we could
> > port the existing one over)
>
>
> If you port the exiting wiki over, you lose independence, no?
>
> Also, I think a wiki is a major overkill. All we need is a static
> webpage that the infrastructure folks can edit, no?
> Sure, that can be implemented by a wiki, but it just seems way way
> more complicated than needed.
why not use google calendar :)
Joshua D. Drake
>
> //Magnus
>
> ---------------------------(end of
> broadcast)--------------------------- TIP 6: explain analyze is your
> friend
>
- --
=== The PostgreSQL Company: Command Prompt, Inc. ===
Sales/Support: +1.503.667.4564 24x7/Emergency: +1.800.492.2240
PostgreSQL solutions since 1997 http://www.commandprompt.com/ UNIQUE NOT NULL
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate
PostgreSQL Replication: http://www.commandprompt.com/products/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQFHOz4VATb/zqfZUUQRAh8qAJ0cW7+sQ+RqZO95QMgygiacqr4IcQCeKk2v
ihqVgRHAID4+Al3NFkV5/I8=
=jsc5
-----END PGP SIGNATURE-----
-----BEGIN PGP SIGNED MESSAGE----- Hash: RIPEMD160 > If you port the exiting wiki over, you lose independence, no? Not really - I meant move the wiki to another completely independent server. > Also, I think a wiki is a major overkill. All we need is a static webpage > that the infrastructure folks can edit, no? > Sure, that can be implemented by a wiki, but it just seems way way more > complicated than needed. A wiki is nice because: * It's dirt simple to edit * It's easy to add and remove write permission to people * There's an automatic history of all changes * It allows for quick organic growth. If we want to break the "Site Status" page into "Website status" and "Mailing listsstatus", it's east to do so. * It has email and rss alerts built in. * It's easier for people to find. I'd love to see all project-related information like this in one place, and a wiki seems like the best way to do so. No reason to make yet another resource to maintain and keep track of. - -- Greg Sabino Mullane greg@turnstep.com End Point Corporation PGP Key: 0x14964AC8 200711141340 http://biglumber.com/x/web?pk=2529DF6AB8F79407E94445B4BC9B906714964AC8 -----BEGIN PGP SIGNATURE----- iD8DBQFHO0EVvJuQZxSWSsgRAw8tAJ9pRRW0XSAqVIpd5vwTNl42sCEOIQCfQjpV F7sMRxLfDn3xFAVauXPaKwc= =cMiQ -----END PGP SIGNATURE-----
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 - --On Wednesday, November 14, 2007 19:23:41 +0100 Magnus Hagander <magnus@hagander.net> wrote: > On Wed, Nov 14, 2007 at 05:42:44PM -0000, Greg Sabino Mullane wrote: >> >> >> > Question is, where should we host that page? >> >> > I'da say there are two options - either on the main website, which is >> > mirrored globally, or on a completely independent system (not just it's own >> > vm, not even in the same datacenter as our other servers). >> >> +1 for independence. Sound to me like a wiki would be perfect for this >> as well. My company would be happy to host a wiki (or we could port the >> existing one over) > > > If you port the exiting wiki over, you lose independence, no? > > Also, I think a wiki is a major overkill. All we need is a static webpage > that the infrastructure folks can edit, no? > Sure, that can be implemented by a wiki, but it just seems way way more > complicated than needed. Why not something that can just be RSS feed into the main site? - ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email . scrappy@hub.org MSN . scrappy@hub.org Yahoo . yscrappy Skype: hub.org ICQ . 7615664 -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4 (FreeBSD) iD8DBQFHO0I44QvfyHIvDvMRAundAJ9XTZrx9fjtElxMoAEx/pFdBVoRjwCg59MS C3IxheETf9BSPnjcGFWxT6g= =WpqQ -----END PGP SIGNATURE-----
Marc G. Fournier wrote: > > > --On Wednesday, November 14, 2007 19:23:41 +0100 Magnus Hagander > <magnus@hagander.net> wrote: > >> On Wed, Nov 14, 2007 at 05:42:44PM -0000, Greg Sabino Mullane wrote: >>> >>>> Question is, where should we host that page? >>>> I'da say there are two options - either on the main website, which is >>>> mirrored globally, or on a completely independent system (not just it's own >>>> vm, not even in the same datacenter as our other servers). >>> +1 for independence. Sound to me like a wiki would be perfect for this >>> as well. My company would be happy to host a wiki (or we could port the >>> existing one over) > >> If you port the exiting wiki over, you lose independence, no? > >> Also, I think a wiki is a major overkill. All we need is a static webpage >> that the infrastructure folks can edit, no? >> Sure, that can be implemented by a wiki, but it just seems way way more >> complicated than needed. > > Why not something that can just be RSS feed into the main site? well it would be fairly easy to drive such a feed from our nagios instance (and even extract stuff like scheduled downtime from it) ... Stefan
Greg Sabino Mullane wrote: > >> If you port the exiting wiki over, you lose independence, no? > > Not really - I meant move the wiki to another completely independent > server. Well, there would be no place to announce wiki outages.. (I'm all for moving the wiki to a dedicated machine, don't get me wrong, but that's a different thing) >> Also, I think a wiki is a major overkill. All we need is a static webpage >> that the infrastructure folks can edit, no? >> Sure, that can be implemented by a wiki, but it just seems way way more >> complicated than needed. > > A wiki is nice because: > > * It's dirt simple to edit Debatable :-P, but yeah, for somethign like this, it's very simple. > * It's easy to add and remove write permission to people Yes. > * There's an automatic history of all changes Check. > * It allows for quick organic growth. If we want to break the "Site Status" > page into "Website status" and "Mailing lists status", it's east to do so. I'd consider that an anti-feature for a site like this, really :-P > * It has email and rss alerts built in. Yes. > * It's easier for people to find. *Huh*? How does the software used determine how easy it is to find a URL? > I'd love to see all project-related information like this in one place, and > a wiki seems like the best way to do so. No reason to make yet another > resource to maintain and keep track of. If we want something to track when things don't work, this shouldn't be in the same system as the things it's supposed to track. Or even on the same server. Or even on the same network. //Magnus
Stefan Kaltenbrunner wrote: > Marc G. Fournier wrote: >> >> --On Wednesday, November 14, 2007 19:23:41 +0100 Magnus Hagander >> <magnus@hagander.net> wrote: >> >>> On Wed, Nov 14, 2007 at 05:42:44PM -0000, Greg Sabino Mullane wrote: >>>>> Question is, where should we host that page? >>>>> I'da say there are two options - either on the main website, which is >>>>> mirrored globally, or on a completely independent system (not just it's own >>>>> vm, not even in the same datacenter as our other servers). >>>> +1 for independence. Sound to me like a wiki would be perfect for this >>>> as well. My company would be happy to host a wiki (or we could port the >>>> existing one over) >>> If you port the exiting wiki over, you lose independence, no? >>> Also, I think a wiki is a major overkill. All we need is a static webpage >>> that the infrastructure folks can edit, no? >>> Sure, that can be implemented by a wiki, but it just seems way way more >>> complicated than needed. >> Why not something that can just be RSS feed into the main site? That's a good idea, IMO. > well it would be fairly easy to drive such a feed from our nagios > instance (and even extract stuff like scheduled downtime from it) ... Not as sure about that one. Basically, not sure we want to publish the automated stuff there, and Nagios really isn't a nice interface to do edits from... JD mentioned google calendar. Can you get an RSS feed from it? If so, that might be a good idea actually - given that it's something that's hosted entirely independent from our current infrastructure. //Magnus
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Wed, 14 Nov 2007 20:44:26 +0100
Magnus Hagander <magnus@hagander.net> wrote:
> Not as sure about that one. Basically, not sure we want to publish the
> automated stuff there, and Nagios really isn't a nice interface to do
> edits from...
>
>
> JD mentioned google calendar. Can you get an RSS feed from it? If so,
> that might be a good idea actually - given that it's something that's
> hosted entirely independent from our current infrastructure.
It looks like it does but I haven't tested it.
Joshua D. Drake
>
> //Magnus
>
> ---------------------------(end of
> broadcast)--------------------------- TIP 6: explain analyze is your
> friend
>
- --
=== The PostgreSQL Company: Command Prompt, Inc. ===
Sales/Support: +1.503.667.4564 24x7/Emergency: +1.800.492.2240
PostgreSQL solutions since 1997 http://www.commandprompt.com/ UNIQUE NOT NULL
Donate to the PostgreSQL Project: http://www.postgresql.org/about/donate
PostgreSQL Replication: http://www.commandprompt.com/products/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
iD8DBQFHO1GjATb/zqfZUUQRAoiMAJ4//hegD+YkZAAU6b9ZI/xfzbICOgCgpFo2
ErB82wzZf3pX8t1YQsX5gPk=
=q3H8
-----END PGP SIGNATURE-----
Magnus Hagander wrote: > Stefan Kaltenbrunner wrote: >> Marc G. Fournier wrote: >>> --On Wednesday, November 14, 2007 19:23:41 +0100 Magnus Hagander >>> <magnus@hagander.net> wrote: >>> >>>> On Wed, Nov 14, 2007 at 05:42:44PM -0000, Greg Sabino Mullane wrote: >>>>>> Question is, where should we host that page? >>>>>> I'da say there are two options - either on the main website, which is >>>>>> mirrored globally, or on a completely independent system (not just it's own >>>>>> vm, not even in the same datacenter as our other servers). >>>>> +1 for independence. Sound to me like a wiki would be perfect for this >>>>> as well. My company would be happy to host a wiki (or we could port the >>>>> existing one over) >>>> If you port the exiting wiki over, you lose independence, no? >>>> Also, I think a wiki is a major overkill. All we need is a static webpage >>>> that the infrastructure folks can edit, no? >>>> Sure, that can be implemented by a wiki, but it just seems way way more >>>> complicated than needed. >>> Why not something that can just be RSS feed into the main site? > > That's a good idea, IMO. not disagreeing here :-) > >> well it would be fairly easy to drive such a feed from our nagios >> instance (and even extract stuff like scheduled downtime from it) ... > > Not as sure about that one. Basically, not sure we want to publish the > automated stuff there, and Nagios really isn't a nice interface to do > edits from... hmm well - there is certainly a lot of stuff on nagios that is probably not appropriate for fully automatic publishing but we could say use the nagios escalation feature for certain services and let that drive the feed. as for editing - nagios has an fairly easy way to set a scheduled downtime for a given host/service (imho more easy than to edit a wiki) but that wont work for stuff not there or special cases ... > > > JD mentioned google calendar. Can you get an RSS feed from it? If so, > that might be a good idea actually - given that it's something that's > hosted entirely independent from our current infrastructure. hmm maybe - but that would be another completely new thing to deal with for admins which I'm not too fond of ... Stefan
Stefan Kaltenbrunner <stefan@kaltenbrunner.cc> writes:
> Magnus Hagander wrote:
>> Stefan Kaltenbrunner wrote:
>>> well it would be fairly easy to drive such a feed from our nagios
>>> instance (and even extract stuff like scheduled downtime from it) ...
>>
>> Not as sure about that one. Basically, not sure we want to publish the
>> automated stuff there, and Nagios really isn't a nice interface to do
>> edits from...
> hmm well - there is certainly a lot of stuff on nagios that is probably
> not appropriate for fully automatic publishing but we could say use the
> nagios escalation feature for certain services and let that drive the feed.
Automated publication of status data on a public website scares me;
it seems like a great way to invite breakins. (Black hat: "whaddya
know, their DNS server is down, maybe I can inject some bogus info.")
I'm for manual entries only on a public-facing page.
regards, tom lane
Tom Lane wrote: > Stefan Kaltenbrunner <stefan@kaltenbrunner.cc> writes: >> Magnus Hagander wrote: >>> Stefan Kaltenbrunner wrote: >>>> well it would be fairly easy to drive such a feed from our nagios >>>> instance (and even extract stuff like scheduled downtime from it) ... >>> Not as sure about that one. Basically, not sure we want to publish the >>> automated stuff there, and Nagios really isn't a nice interface to do >>> edits from... > >> hmm well - there is certainly a lot of stuff on nagios that is probably >> not appropriate for fully automatic publishing but we could say use the >> nagios escalation feature for certain services and let that drive the feed. > > Automated publication of status data on a public website scares me; > it seems like a great way to invite breakins. (Black hat: "whaddya > know, their DNS server is down, maybe I can inject some bogus info.") well that would only include stuff that is publically available(or rather a public facing service) anyway (there is nothing that stops somebody to check the availability of say our DNS-servers or say of wwwmaster by himself). There is at least one precedence for doing this too: http://monitoring.apache.org/status/ > > I'm for manual entries only on a public-facing page. fair enough - I'm just not too happy about having too many things one has to deal with in such a case (updating a wiki, scheduling maintainance in nagios to avoid people getting alerted, send mail to -www and -hackers, ...). Too complex procedures will hurt and not encourage all involved people to handle planned stuff in the way it should ... Stefan
On Tue, Nov 13, 2007 at 09:39:44AM -0800, Josh Berkus wrote: > Marc, > > > I put out an announcement several days ago that I would be doing > > work on neptune that would cause some disruptions to clients on > > Saturday (ie. extended downtime) ... Magnus confirmed with me > > that this wouldn't affect any of the Pg VPSs, but I wasn't clear > > in that although there would be no extended downtime for the Pg > > VPSs, that I still had to move the affected Pg VPSs *off* of the > > server being worked on ... my fault, I should have been more clear > > on that point ... > > It would be nice to have a "planned outages" page somewhere which > people could check. It would save us a bunch of answering e-mail, > if nothing else. And the outage announcements could get carred in > PWN if Fetter is amenable. I'd be delighted :) Cheers, David. -- David Fetter <david@fetter.org> http://fetter.org/ Phone: +1 415 235 3778 AIM: dfetter666 Yahoo!: dfetter Skype: davidfetter XMPP: david.fetter@gmail.com Remember to vote! Consider donating to Postgres: http://www.postgresql.org/about/donate
-----BEGIN PGP SIGNED MESSAGE----- Hash: RIPEMD160 > Well, there would be no place to announce wiki outages.. Well, true, but if not the wiki, where would we announce outages to the outages list? :) I presume known upcoming wiki outages would be announced on the wiki, and actual downtime announced on the main page or lists. In actuality, I think wiki outages would be a very rare event. > (I'm all for moving the wiki to a dedicated machine, don't get me wrong, > but that's a different thing) Agree with part of this: see below. >> * It's easier for people to find. > *Huh*? > How does the software used determine how easy it is to find a URL? I meant as far as one less URL for people to know about and track, rather than having yet another Postgres resource to feed and feed and publicize. > If we want something to track when things don't work, this shouldn't be > in the same system as the things it's supposed to track. Or even on the > same server. Or even on the same network. Agreed to a point, but I don't know that tracking the wiki is really that important when compared to the other things: web site, mirrors, cvs, lists, and bug tracker. (Just kidding on that last one.) Ideally, we'd insulate things even more widely than we have them right now, but we're moving in the right direction the last few years. Heck, we could even look at putting the wiki on two servers, with a database and web server on each one, with DNS pointing to each randomly. Could be a cool experiment. - -- Greg Sabino Mullane greg@turnstep.com End Point Corporation PGP Key: 0x14964AC8 200711151131 http://biglumber.com/x/web?pk=2529DF6AB8F79407E94445B4BC9B906714964AC8 -----BEGIN PGP SIGNATURE----- iD8DBQFHPHShvJuQZxSWSsgRA8sYAKCSx05lBht7uteI/2HhnA6Pri8DlwCfQXBY B6BEaPfe+yMM6KqxCG9tui0= =pnx+ -----END PGP SIGNATURE-----
Greg Sabino Mullane wrote: > >> Well, there would be no place to announce wiki outages.. > > Well, true, but if not the wiki, where would we announce outages to > the outages list? :) I presume known upcoming wiki outages would be > announced on the wiki, and actual downtime announced on the main page > or lists. In actuality, I think wiki outages would be a very rare event. *All* our outages should be rare events. Doesn't invalidate the point. >>> * It's easier for people to find. > >> *Huh*? >> How does the software used determine how easy it is to find a URL? > > I meant as far as one less URL for people to know about and track, rather > than having yet another Postgres resource to feed and feed and publicize. I think it's better to have a dedicated place - where project folks can go. Then we feed it with RSS to the main site for "outsiders". >> If we want something to track when things don't work, this shouldn't be >> in the same system as the things it's supposed to track. Or even on the >> same server. Or even on the same network. > > Agreed to a point, but I don't know that tracking the wiki is really that > important when compared to the other things: web site, mirrors, cvs, lists, > and bug tracker. (Just kidding on that last one.) Ideally, we'd insulate things > even more widely than we have them right now, but we're moving in the right > direction the last few years. > > Heck, we could even look at putting the wiki on two servers, with a database > and web server on each one, with DNS pointing to each randomly. Could be a > cool experiment. > Seems like an experiment is not what we want for this :-P //Magnus
On Wed, Nov 14, 2007 at 03:13:58PM -0500, Tom Lane wrote: > it seems like a great way to invite breakins. (Black hat: "whaddya > know, their DNS server is down, maybe I can inject some bogus info.") If we ever have a complete DNS outage, then we need more DNS servers. Which reminds me, I had promised to try to get postgresql.org on the afilias name server infrastructure (which, I assure you, had better _not_ ever be down). Are people still interested in that? A -- Andrew Sullivan Old sigs will return after re-constitution of blue smoke
On Wed, Nov 14, 2007 at 09:32:04PM +0100, Stefan Kaltenbrunner wrote: > has to deal with in such a case (updating a wiki, scheduling > maintainance in nagios to avoid people getting alerted, send mail to > -www and -hackers, ...). I think there is something to be said for the above point. We don't want 10 minutes of emergency-panic-coping before actually coping with the emergency. This is why I like the idea of a way to publish certain alerts automatically. (I don't think every time Nagios goes "bing" it oughta show up on the website, but I don't think anyone was suggesting that either.) A -- Andrew Sullivan Old sigs will return after re-constitution of blue smoke
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 - --On Friday, November 16, 2007 12:48:40 -0500 Andrew Sullivan <ajs@crankycanuck.ca> wrote: > On Wed, Nov 14, 2007 at 03:13:58PM -0500, Tom Lane wrote: >> it seems like a great way to invite breakins. (Black hat: "whaddya >> know, their DNS server is down, maybe I can inject some bogus info.") > > If we ever have a complete DNS outage, then we need more DNS servers. Which > reminds me, I had promised to try to get postgresql.org on the afilias name > server infrastructure (which, I assure you, had better _not_ ever be down). > Are people still interested in that? No, but thanks ... if Afilias could provide another backup, that would be cool for redundancy purposes ... - ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email . scrappy@hub.org MSN . scrappy@hub.org Yahoo . yscrappy Skype: hub.org ICQ . 7615664 -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4 (FreeBSD) iD8DBQFHPdk64QvfyHIvDvMRAiSTAKCR2yAmrZK9i14n24LLO55XrcLIQwCcDF8E v+XVdXMXQfbNfARvtzamuh0= =sFqs -----END PGP SIGNATURE-----
Andrew Sullivan wrote: > On Wed, Nov 14, 2007 at 03:13:58PM -0500, Tom Lane wrote: >> it seems like a great way to invite breakins. (Black hat: "whaddya >> know, their DNS server is down, maybe I can inject some bogus info.") > > If we ever have a complete DNS outage, then we need more DNS servers. Which > reminds me, I had promised to try to get postgresql.org on the afilias name > server infrastructure (which, I assure you, had better _not_ ever be down). > Are people still interested in that? What does that actually mean? A well-distributed and always-up secondary? Or the master as well, including a new management system? //Magnus
On Fri, Nov 16, 2007 at 01:54:02PM -0400, Marc G. Fournier wrote: > > No, but thanks ... if Afilias could provide another backup, that would be cool > for redundancy purposes ... Right, it wasn't a wholesale replacement, but another set of nameservers. We have currently five large nodes anycasted around the world, serving from Amsterdam, Miami, Toronto, Seattle, and (in process of turn up this week) Hong Kong. If we added one or two nameservers to the NS set, then we should get coverage from everywhere in the world. So if people still want it, I'll see what I can do. (There are some rules about which IPs are allowed to be used, so I'll have to make sure we can do it, but I can't see why not.) A -- Andrew Sullivan Old sigs will return after re-constitution of blue smoke
On Fri, Nov 16, 2007 at 07:05:12PM +0100, Magnus Hagander wrote: > > What does that actually mean? > > A well-distributed and always-up secondary? Or the master as well, > including a new management system? The former was what I'd offered to get permission for before. A -- Andrew Sullivan Old sigs will return after re-constitution of blue smoke
I could provide a slave NS as well.
On 11/16/07, Andrew Sullivan <ajs@crankycanuck.ca> wrote:
On Fri, Nov 16, 2007 at 01:54:02PM -0400, Marc G. Fournier wrote:
>
> No, but thanks ... if Afilias could provide another backup, that would be cool
> for redundancy purposes ...
Right, it wasn't a wholesale replacement, but another set of nameservers.
We have currently five large nodes anycasted around the world, serving from
Amsterdam, Miami, Toronto, Seattle, and (in process of turn up this week)
Hong Kong. If we added one or two nameservers to the NS set, then we should
get coverage from everywhere in the world. So if people still want it, I'll
see what I can do. (There are some rules about which IPs are allowed to be
used, so I'll have to make sure we can do it, but I can't see why not.)
A
--
Andrew Sullivan
Old sigs will return after re-constitution of blue smoke
---------------------------(end of broadcast)---------------------------
TIP 2: Don't 'kill -9' the postmaster
Andrew Sullivan wrote: > On Fri, Nov 16, 2007 at 07:05:12PM +0100, Magnus Hagander wrote: >> What does that actually mean? >> >> A well-distributed and always-up secondary? Or the master as well, >> including a new management system? > > The former was what I'd offered to get permission for before. Then my vote is yes, please. I find it unlikely that we'll find a better DNS provider doing it for us for free :-) //Magnus
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 - --On Friday, November 16, 2007 13:41:15 -0500 Andrew Sullivan <ajs@crankycanuck.ca> wrote: > On Fri, Nov 16, 2007 at 01:54:02PM -0400, Marc G. Fournier wrote: >> >> No, but thanks ... if Afilias could provide another backup, that would be >> cool for redundancy purposes ... > > Right, it wasn't a wholesale replacement, but another set of nameservers. > We have currently five large nodes anycasted around the world, serving from > Amsterdam, Miami, Toronto, Seattle, and (in process of turn up this week) > Hong Kong. If we added one or two nameservers to the NS set, then we should > get coverage from everywhere in the world. So if people still want it Sounds perfect, thank you ... just let me know which IPs to allow xfer's from, andwhat to add to the domain record ... - ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email . scrappy@hub.org MSN . scrappy@hub.org Yahoo . yscrappy Skype: hub.org ICQ . 7615664 -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4 (FreeBSD) iD8DBQFHPeXe4QvfyHIvDvMRAhlvAKDBOyvfUKq1e+pazLT1lTbkz87l2wCgr/RZ lTfcjYcqzgEgEywjLXAl+h0= =oold -----END PGP SIGNATURE-----
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 - --On Friday, November 16, 2007 13:42:50 -0500 "Gavin M. Roy" <gmr@ehpg.net> wrote: > I could provide a slave NS as well. Sounds cool to me ... from what I can tell, a .org is allowed, what, 13 name servers? Just let me know IP(s) to add, and I'll make the change ... Thanks ... - ---- Marc G. Fournier Hub.Org Networking Services (http://www.hub.org) Email . scrappy@hub.org MSN . scrappy@hub.org Yahoo . yscrappy Skype: hub.org ICQ . 7615664 -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.4 (FreeBSD) iD8DBQFHPeZn4QvfyHIvDvMRAmCSAJ4u4xqCYuKCsAHexcseGNYLlB2ZLQCgj4F2 3d2F7RQfjNfdwHdJQmZ/5ss= =thTV -----END PGP SIGNATURE-----
On Fri, Nov 16, 2007 at 02:50:15PM -0400, Marc G. Fournier wrote: > > I could provide a slave NS as well. > > Sounds cool to me ... from what I can tell, a .org is allowed, what, 13 name > servers? Just let me know IP(s) to add, and I'll make the change ... Nobody is allowed more than 13 name servers, because 13 is the rule of thumb for when the answer will go over the protocol limit and cause fallback to TCP. (Actually, these days, it'll probably do EDNS0 instead, but whatever.) Anyway, the evidence seems to be that somewhere around 7 or 8, there's no more benefit. After that, one usually finds that some of the servers never get any traffic at all, because most DNS servers are robust enough not to go down very often, and recursive resolvers remember who gave them the answer last time, and usually keep going there. A -- Andrew Sullivan Old sigs will return after re-constitution of blue smoke