On Sunday 06 April 2003 05:00 am, Daniel Ehrenberg wrote:
aparently, the wikipedia server is down. This hasn't been reported on the list, although this might have been because the email server is down too.
You might be surprised to know that the Wikipedia encyclopedia and the Wikipedia mail server are both on the same box.
Could all the mailing lists work just fine on a separate cheapo AMD K6-II 500 MHZ Linux box with two 4 gig hard drives, 256MB of SDRAM and a couple network cards? If so, I hereby donate such a dinosaur to serve as a mail server (of course I'll have to test these components to see if they still work so the specs may change - somebody else will have to configure the mail server part though).
Not being able to talk to each other as a group during a crash like this isn't exactly an optimal way to run a project the size of Wiki(p/m)edia.
-- Daniel Mayer (aka mav)
WikiKarma The usual at [[March 29]]
Daniel Mayer wrote:
Could all the mailing lists work just fine on a separate cheapo AMD K6-II 500 MHZ Linux box with two 4 gig hard drives, 256MB of SDRAM and a couple network cards? If so, I hereby donate such a dinosaur to serve as a mail server (of course I'll have to test these components to see if they still work so the specs may change - somebody else will have to configure the mail server part though).
I think every Pentium 1 @133MHz can run the mailing lists. This host you describe can handle more.
Smurf
(Daniel Mayer maveric149@yahoo.com):
Could all the mailing lists work just fine on a separate cheapo AMD K6-II 500 MHZ Linux box...
I'm sure hardware isn't the problem, just convenience. But it might not be a bad idea for maybe the wikitech list and possibly the wikidown address to be run on a different host, maybe even a different colo site (like at Piclab or even some freemail service).
Halloechen!
Lee Daniel Crocker lee@piclab.com writes:
(Daniel Mayer maveric149@yahoo.com):
Could all the mailing lists work just fine on a separate cheapo AMD K6-II 500 MHZ Linux box...
I'm sure hardware isn't the problem, just convenience. But it might not be a bad idea for maybe the wikitech list and possibly the wikidown address to be run on a different host, maybe even a different colo site (like at Piclab or even some freemail service).
Maybe on Wikipedia's Sourceforge project page?
Tschoe, Torsten.
On Sun, 2003-04-06 at 07:02, Daniel Mayer wrote:
Could all the mailing lists work just fine on a separate cheapo AMD K6-II 500 MHZ Linux box with two 4 gig hard drives, 256MB of SDRAM and a couple network cards? If so, I hereby donate such a dinosaur to serve as a mail server
Yes, that would be more than enough to host the mail (and perhaps the backup dumps?). Best talk to Jimbo and Jason about transporting and setting up hardware.
As Lee has suggested, it might also be good to have some stuff at a separate provider in case the hosting is knocked out altogether.
Not being able to talk to each other as a group during a crash like this isn't exactly an optimal way to run a project the size of Wiki(p/m)edia.
What, you didn't enjoy your nice relaxing 22-hour break from routine? :)
-- brion vibber (brion @ pobox.com)
Moving the mailing lists to a different machine would be fairly easy to do. The only troubles that might arise (that I foresee) are these:
* We have limited rack space in the colo. All of our current servers are either 1U or 2U rackmount servers. A standard tower-type server would be a bit of a squeeze in there...
* we are almost out of IP addresses in our current block of allocated IPs, I don't recall how easy it is for us to have another block routed to our cage.
* our remote power control unit is nearly full. This is currently our only means of rebooting a machin that has lost internet connectivity (aside from a 200 mile drive by your truly). Of course, this isn't necessarily a deal-breaker, just a thought. If anybody has a IP based KVM switch available... :)
* come to think of power, I'm not sure how many watts we have going into that cage either. Of course, this is also something that can be fixed if our current setup doesn't have enough power.
That's my piece...
Jason Richey
Daniel Mayer wrote:
On Sunday 06 April 2003 05:00 am, Daniel Ehrenberg wrote:
aparently, the wikipedia server is down. This hasn't been reported on the list, although this might have been because the email server is down too.
You might be surprised to know that the Wikipedia encyclopedia and the Wikipedia mail server are both on the same box.
Could all the mailing lists work just fine on a separate cheapo AMD K6-II 500 MHZ Linux box with two 4 gig hard drives, 256MB of SDRAM and a couple network cards? If so, I hereby donate such a dinosaur to serve as a mail server (of course I'll have to test these components to see if they still work so the specs may change - somebody else will have to configure the mail server part though).
Not being able to talk to each other as a group during a crash like this isn't exactly an optimal way to run a project the size of Wiki(p/m)edia.
-- Daniel Mayer (aka mav)
WikiKarma The usual at [[March 29]] _______________________________________________ Wikitech-l mailing list Wikitech-l@wikipedia.org http://www.wikipedia.org/mailman/listinfo/wikitech-l
On Mon, Apr 07, 2003 at 12:40:12AM -0700, Jason Richey wrote:
Moving the mailing lists to a different machine would be fairly easy to do. The only troubles that might arise (that I foresee) are these:
We have limited rack space in the colo. All of our current servers are either 1U or 2U rackmount servers. A standard tower-type server would be a bit of a squeeze in there...
we are almost out of IP addresses in our current block of allocated IPs, I don't recall how easy it is for us to have another block routed to our cage.
our remote power control unit is nearly full. This is currently our only means of rebooting a machin that has lost internet connectivity (aside from a 200 mile drive by your truly). Of course, this isn't necessarily a deal-breaker, just a thought. If anybody has a IP based KVM switch available... :)
come to think of power, I'm not sure how many watts we have going into that cage either. Of course, this is also something that can be fixed if our current setup doesn't have enough power.
That's my piece...
Jason Richey
Daniel Mayer wrote:
On Sunday 06 April 2003 05:00 am, Daniel Ehrenberg wrote:
aparently, the wikipedia server is down. This hasn't been reported on the list, although this might have been because the email server is down too.
You might be surprised to know that the Wikipedia encyclopedia and the Wikipedia mail server are both on the same box.
Could all the mailing lists work just fine on a separate cheapo AMD K6-II 500 MHZ Linux box with two 4 gig hard drives, 256MB of SDRAM and a couple network cards? If so, I hereby donate such a dinosaur to serve as a mail server (of course I'll have to test these components to see if they still work so the specs may change - somebody else will have to configure the mail server part though).
Not being able to talk to each other as a group during a crash like this isn't exactly an optimal way to run a project the size of Wiki(p/m)edia.
-- Daniel Mayer (aka mav)
I would be willing to host the mailing lists on my server (www.twoevils.org). It is a 2x PII 233MHz, 256MB RAM, with about 40GB of disk space. I'm running Postfix, so it would have to work with that, but otherwise, I don't have any problems with it. It is connected to a 1Mbit DSL line, so it has quite a bit of bandwidth.
On Mon, Apr 07, 2003 at 12:40:12AM -0700, Jason Richey wrote:
We have limited rack space in the colo. All of our current servers are either 1U or 2U rackmount servers. A standard tower-type server would be a bit of a squeeze in there...
we are almost out of IP addresses in our current block of allocated IPs, I don't recall how easy it is for us to have another block routed to our cage.
our remote power control unit is nearly full. This is currently our only means of rebooting a machin that has lost internet connectivity (aside from a 200 mile drive by your truly). Of course, this isn't necessarily a deal-breaker, just a thought. If anybody has a IP based KVM switch available... :)
Linux can be set up to use a serial console. The most PCs still have two serial ports. If you assign the console to port A, port B can be used to access the console of another server. Having two servers would allow us to set them up in this way. If the web server hangs with "Give root password for system maintenance mode" you can connect to the server using e.g. minicom and fix it.
High-End solution would be the use of a terminal server, but this is much more expensive, uses rack space and require an additional IP.
KVM's are for windows users ... :-)
Regards,
JeLuF
That's a good point... The primary problem is that our servers (or most of them, anyway) have video on-board, and only one serial port (the video port takes the place of the second one). I'm sure the motherboards have internal connections for an extra serial port (actually, I'm not quite "sure"), but we haven't looked very far into going that route.
Add all that to the fact that I've never set up a serial console on Linux, and I don't even remember how to make a null modem cable (of course, I still remember how to BUY one), and you can see why we haven't tried that yet.
I'm always willing to listen if someone can point me to a refresher course...
Jason
Jens Frank wrote:
On Mon, Apr 07, 2003 at 12:40:12AM -0700, Jason Richey wrote:
We have limited rack space in the colo. All of our current servers are either 1U or 2U rackmount servers. A standard tower-type server would be a bit of a squeeze in there...
we are almost out of IP addresses in our current block of allocated IPs, I don't recall how easy it is for us to have another block routed to our cage.
our remote power control unit is nearly full. This is currently our only means of rebooting a machin that has lost internet connectivity (aside from a 200 mile drive by your truly). Of course, this isn't necessarily a deal-breaker, just a thought. If anybody has a IP based KVM switch available... :)
Linux can be set up to use a serial console. The most PCs still have two serial ports. If you assign the console to port A, port B can be used to access the console of another server. Having two servers would allow us to set them up in this way. If the web server hangs with "Give root password for system maintenance mode" you can connect to the server using e.g. minicom and fix it.
High-End solution would be the use of a terminal server, but this is much more expensive, uses rack space and require an additional IP.
KVM's are for windows users ... :-)
Regards,
JeLuF
Wikitech-l mailing list Wikitech-l@wikipedia.org http://www.wikipedia.org/mailman/listinfo/wikitech-l
wikitech-l@lists.wikimedia.org