Re: [Server-devel] XSCE install on cubox

2014-04-22 Thread John Watlington

George, thanks for doing this!

On Apr 22, 2014, at 11:48 AM, George Hunt wrote:

 Just a small step, but XSCE 5.0 now installs on Cubox. Start with Tim's 
 image, and comment out ajenti in roles/core/meta/main.yml. I re-rolled 
 xs-moodle, ds-backup to use cronie rather than vixie-cron, and PyYAML rather 
 than python-syck.
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


[Server-devel] XMPP

2014-04-22 Thread John Watlington

Just a quick question: has anybody gotten the XOs to collaborate
properly with an XMPP server other than ejabberd ?

wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] constant power add-tag CP fails on XO-4?

2014-01-24 Thread John Watlington

On Jan 24, 2014, at 6:36 PM, Adam Holt wrote:

 We've relied on this for many years on XO-1s and XO-1.75s (etc) to ensure 
 small XO servers auto-boot after the inevitable power failures -- and yet 
 today it (apparently) no longer works.
 
 Is it possible this does not work on XO-4s, or are we somehow entering 
 add-tag CP incorrectly at the ok prompt, on this XO-4 Touch SKU306 (with 
 the latest stable firmware Q7B37 below) in Haiti?

I just tested on an XO-4 with Q7B37, and behavior with the CP tag set seems OK.

I'm sure you've used .mfg-data to make sure the tag is being set properly,
so perhaps this is a unit-specific hardware bug ?   Any chance of trying 
another laptop ?

Cheers,
wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Root fs on XO1

2013-08-09 Thread John Watlington

Just to clarify a comment that James made:

On XO-1, the qualification of the SD card interface was done by Quanta.
They seemed to think it worked fine.   When I started doing extensive
testing of SD cards in preparation for using them in XO-1.5, I discovered
that some XO-1s did not have a reliable SD card interface.Bit error
rates aren't high (maybe an error every few days when continuously
exercising the card), but unacceptable from my point of view.

So as James said, it might work for you, but not work so well on an
arbitrary XO-1.

On later XO models, I personally tested the SD interface(s) with a
wide variety of SD cards to avoid repeating the same mistake.

Regards,
wad


On Aug 9, 2013, at 9:31 AM, Mikus Grinbergs wrote:

 On 08/09/2013 06:17 AM, James Cameron wrote:
 I have never been happy with using the
 XO-1 SD card slot.
 
 I have been using SD cards in the XO-1 SD card slot for more than five years 
 now.  Although I have experienced occasions of SD card corruption, they have 
 been so rare as to not affect what I've been doing with my XO-1 systems (I 
 just clean the SD card and keep on using it).  In those five years I have 
 had maybe five SD card failures (the SD card stops responding electrically) 
 out of a pool of about 40 cards -- I consider my SD cards to have provided me 
 acceptable reliability. [Contrast that with my experience with XO-1.5 
 systems - four out of ten failed (admittedly, the failures were early 
 systems).]
 
 By providing a swap partition on the SD card, I've been able to run *large* 
 Linux applications (e.g., BOINC, gvSIG) on the XO-1, despite its limited main 
 memory (I run them from Terminal in the Sugar environment, and live with the 
 limited multi-window capability provided by Sugar). What I place on the SD 
 card is executables (e.g., Adobe, Java, Browsers, Sugar Activities (3GB+), 
 Timidity) and data (mainly accessed through Terminal - Movies, Books, Music, 
 Images, Maps, etc.).
 
 Without my 'permanent' SD card. the XO-1 would be too little for me.
 
 mikus
 
 
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] School networks and electrical equipment damage

2013-06-06 Thread John Watlington

On Jun 6, 2013, at 3:58 PM, Daniel Drake wrote:

 Those of us familiar with setting up school networks (server + switch
 + APs) in some of our deployments will be familiar with  the
 occasional loss of hardware, due to surges in the low quality
 electrical supply or whatever, even when the system is protected by a
 cheap UPS which supposedly offers some protection.
 
 This has often been the case in Nicaragua, so the group is now buying
 more expensive UPSes, PoE switches, and PoE access points for new
 schools. This means that the server and switch are connected to mains
 power via a UPS which hopefully protects them, and none of the APs are
 connected directly to the mains (instead they get Power over Ethernet)
 which hopefully offers some isolation from bad electrical conditions.
 
 This equipment is expensive, especially in places like Nicaragua where
 lots of import taxes are applied. But the hope is that the investment
 pays off in that the equipment doesn't get zapped.
 
 However, one week after deploying this equipment in the first school,
 we are left with a server that doesn't boot, 3 out of 4 access points
 broken with a nice burning electronics smell, and a broken switch with
 a lot of visible damage to the electronics.
 
 And the most surprising thing - we had not even turned on the network
 yet, pending some electrical work. Everything was connected up except
 one crucial link - the UPS was not plugged into mains power. So all of
 this damage happened without any of the devices having a connection to
 the mains.
 
 Connectivity-wise, the setup was:
 WAN: Phone line - ADSL modem - XS
 LAN: XS - Switch - 4 APs
 
 And power connections: the XS, ADSL modem and switch were connected to
 the UPS. The APs were connected to the switch over ethernet for both
 power and data. Again, since the battery was not connected to mains
 power, none of the devices had a power source.
 
 The connectivity engineer's best bet is that a lightening bolt landed
 at the school or nearby, and that this caused a power surge on the
 phone line. This surge passed through the ADSL modem, server, switch,
 and 4 APs, destroying everything in its path (except 1 AP that was
 connected over a longer cable than the rest).

This was my most likely hypothesis as well.
I believe the damage would have been less had the UPS
actually been plugged in, but most probably have their input
protected, not their outputs!

 I figured this is a story worth sharing, for any other projects
 considering splashing out on more expensive equipment...

 Also, I'm wondering if anyone has any advice/experience here. Would
 others expect this more expensive setup to be more resilient to bad
 electrical conditions than a cheaper setup - will the investment pay
 off?

Cat5 Ethernet transformers generally provide 1.5 kV of isolation
(at each end) but PoE breaks that.

I believe the actual protection provided by the UPS can vary widely.
They are usually required in situations like yours to protect the hard
drivers, not as much for power line protection.   I would use a periodically
replaced surge protector before it and maybe a surge protected power strip
after it.

 I figure that the case of a lightening bolt might be a bit extreme,
 but electrical storms are a nightly occurance here almost daily during
 the 6 month rainy season.

Nothing will protect against a direct lightning strike of the wire.
The more common case is a strike near the wire or the central office which
can induce still induce many kV of surge on the lines.   If electrical storms
are common you need to take precautions.

I hope all the network cabling is indoors ?   If it is only partially so,
consider something like:
http://www.l-com.com/surge-protector-indoor-4-port-med-power-10-100-base-t-cat5-lightning-surge-protector

 I have seen that some UPSs (unfortunately not these ones) allow a
 phone line to be passed through them, supposedly offering some
 protection. Would such a system protect against a lightening bolt,
 assuming thats what happened here?

Probably not.

Primary protection on the phone line should be an gapped carbon block
or gas tube protector at the entrance to the building between each line of
the phone pair and a good ground.These protect against higher
energy surges, and generally kick in at over 1 kV.

I would suggest using primary protection that includes secondary protection,
such as you see in:
http://www.l-com.com/surge-protector-outdoor-high-power-telephone-dsl-lightning-surge-protector-screw-terminals

You could also loop it through the UPS at this point for redundancy
(but it MUST remain plugged in to provide protection --- it not, it makes
things worse!)

Then there is the protection in the modem itself, which should be able
to handle the remaining surge.   They are so common in telephony as to be
required --- but primary protection and proper grounding is always assumed.

Sorry to hear about your misfortunes,
wad


Re: [Server-devel] failed to register

2012-06-04 Thread John Watlington

On Jun 4, 2012, at 7:30 PM, John Watlington wrote:

 Adam, something is wrong with those addresses.
 The 18.nn.nn.nn subnet is owned by MIT.   You shouldn't be
 using any of them.Perhaps you meant 172.18.nn.nn ?

And I never got around to the second part of the comment.
eth0 (the WAN port) on the XS should be in a different subnet
than the laptops (the LAN ports).If for some reason it is not,
difficulties will ensue as the routing tables will be screwed up.

Cheers,
wad


 On Jun 4, 2012, at 6:19 PM, Holt wrote:
 
 On 6/4/2012 3:27 PM, Sameer Verma wrote:
 On Sun, Jun 3, 2012 at 3:24 PM, Jerry Vonaujvo...@shaw.ca  wrote:
 On Sat, 2012-06-02 at 13:15 -0700, Sameer Verma wrote:
 On Sat, Jun 2, 2012 at 10:10 AM, Sameer Vermasve...@sfsu.edu  wrote:
 XS 0.7
 OLPC build 874
 Sugar 0.92.2
 
 The XO fails to register, but a folder is created for the XO on the XS
 under /library/users/
 
 The XO shows no collaboration server under Control Panel | Network
 
 Any ideas? Pointers?
 
 cheers,
 Sameer
 On the XS I see the machine as registered via /var/log/messages
 (schoolserver olpc_idmgr: Registered user abcdefgh with serial
 SHC03345845)
 
 So, looks like the server end is doing what it should, but the XO end is 
 not?
 Think you bumped into this: https://dev.laptop.org/ticket/11056
 
 Jerry
 
 Thx. Will investigate. Copying Adam Holt and the Jamaica team on it
 (the trouble in in a school there).
 
 Sameer
 
 One XO-1.5 successfully registered back on Saturday.  It successfully then 
 pushed 36MB to the XS' /library/users/SHC03801C2E (after running 
 /usr/bin/ds-backup.sh on the XO-1.5 and waiting ~30min).  So we went home 
 Saturday night with a false sense of confidence!
 
 But no XO-1.5s will fully register today (all ~50 of the school's XOs are 
 XO-1.5s).
 Clearly, Jamaica's change from XS 0.6 to XS 0.7 two months ago destroyed 
 reliable registration?
 
 Frustrating as all XO-1.5s do connect to XS, whose /library/users become 
 populated with each XO's serial number, when they 1st attempt to register.  
 EG. directory /library/users/SHC03801C18 on XS is just 1 of many examples of 
 an XO-1.5 incompletely registering -- all similar directories contain:
 
   .bash_logout
   .bash_profile
   .bashrc
   .ssh
 
 Is it normal that XO-1.5s can ping the schoolserver in these 3 ways:
 
   ping 18.172.0.1
   ping schoolserver(responds as 18.172.196.1)
   ping schoolserver.providence.uwimona.edu.jm(responds as 18.172.196.1)
 
 But yet CANNOT ping the schoolserver's eth0 IP address:
 
   ping 18.172.196.1(its inet addr according to ifconfig etho anyway)
 
 If it matters, a failng-to-register XO-1.5's /etc/resolv.conf reads as 
 follows:
 
   # Generated by NetworkManager
   domain providence.uwimona.edu.jm
   search providence.uwimona.edu.jm localdomain
   nameserver 172.18.196.1
 
 What should I be trying?? :)
 
 --
 Help kids everywhere map their world, at http://olpcMAP.net !
 
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel
 
 

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] failed to register

2012-06-04 Thread John Watlington

On Jun 4, 2012, at 8:35 PM, John Watlington wrote:

 I suspect that your ISP is assigning 172.18.196.1 to the
 XS, which is then confused as to where packets destined
 for 172.18.xx.xx should be routed.
 eth0 and the other interfaces on the XS should be in separate
 subnets.

BTW, if this is actually the problem I don't have a quick fix.
This is why I chose the relatively unused 172.18.xx.xx unroutable
addresses originally, instead of the more common 10.xx.xx.xx or
192.168.xx.xx private ranges for the internal network.
Perhaps someone who has worked on the XS networking recently
can help ?

wad

 On Jun 4, 2012, at 8:04 PM, Holt wrote:
 
 On 6/4/2012 7:32 PM, John Watlington wrote:
 On Jun 4, 2012, at 7:30 PM, John Watlington wrote:
 
 Adam, something is wrong with those addresses.
 The 18.nn.nn.nn subnet is owned by MIT.   You shouldn't be
 using any of them.Perhaps you meant 172.18.nn.nn ?
 
 Wad's exactly right -- I spent too many years at MIT and so typed 18.* 
 without thinking.  The correct IPs for the XS (no idea why there are 2) are 
 in fact:
   172.18.0.1  (pingable, ssh, etc)
   172.18.196.1  (schoolserver?)
 
 I remain very concerned that 172.18.196.1 is not pingable from the XO-1.5s, 
 even when this IP's names {schoolserver, 
 schoolserver.providence.uwimona.edu.jm} are pingable, responding with 
 172.18.196.1
 
 In Response to DSD:
 There is no server set in My Settings  Network  Server on any of the 
 XO-1.5s
 
 Hopefully Sameer who helped with the original XS networking config 2 months 
 ago can respond to Wad's WAN/LAN request below??
 
 And I never got around to the second part of the comment.
 eth0 (the WAN port) on the XS should be in a different subnet
 than the laptops (the LAN ports).If for some reason it is not,
 difficulties will ensue as the routing tables will be screwed up.
 
 Cheers,
 wad
 
 
 On Jun 4, 2012, at 6:19 PM, Holt wrote:
 
 On 6/4/2012 3:27 PM, Sameer Verma wrote:
 On Sun, Jun 3, 2012 at 3:24 PM, Jerry Vonaujvo...@shaw.ca   wrote:
 On Sat, 2012-06-02 at 13:15 -0700, Sameer Verma wrote:
 On Sat, Jun 2, 2012 at 10:10 AM, Sameer Vermasve...@sfsu.edu   wrote:
 XS 0.7
 OLPC build 874
 Sugar 0.92.2
 
 The XO fails to register, but a folder is created for the XO on the XS
 under /library/users/
 
 The XO shows no collaboration server under Control Panel | Network
 
 Any ideas? Pointers?
 
 cheers,
 Sameer
 On the XS I see the machine as registered via /var/log/messages
 (schoolserver olpc_idmgr: Registered user abcdefgh with serial
 SHC03345845)
 
 So, looks like the server end is doing what it should, but the XO end 
 is not?
 Think you bumped into this: https://dev.laptop.org/ticket/11056
 
 Jerry
 
 Thx. Will investigate. Copying Adam Holt and the Jamaica team on it
 (the trouble in in a school there).
 
 Sameer
 One XO-1.5 successfully registered back on Saturday.  It successfully 
 then pushed 36MB to the XS' /library/users/SHC03801C2E (after running 
 /usr/bin/ds-backup.sh on the XO-1.5 and waiting ~30min).  So we went home 
 Saturday night with a false sense of confidence!
 
 But no XO-1.5s will fully register today (all ~50 of the school's XOs are 
 XO-1.5s).
 Clearly, Jamaica's change from XS 0.6 to XS 0.7 two months ago destroyed 
 reliable registration?
 
 Frustrating as all XO-1.5s do connect to XS, whose /library/users become 
 populated with each XO's serial number, when they 1st attempt to 
 register.  EG. directory /library/users/SHC03801C18 on XS is just 1 of 
 many examples of an XO-1.5 incompletely registering -- all similar 
 directories contain:
 
  .bash_logout
  .bash_profile
  .bashrc
  .ssh
 
 Is it normal that XO-1.5s can ping the schoolserver in these 3 ways:
 
  ping 18.172.0.1
  ping schoolserver(responds as 18.172.196.1)
  ping schoolserver.providence.uwimona.edu.jm(responds as 18.172.196.1)
 
 But yet CANNOT ping the schoolserver's eth0 IP address:
 
  ping 18.172.196.1(its inet addr according to ifconfig etho 
 anyway)
 
 If it matters, a failng-to-register XO-1.5's /etc/resolv.conf reads as 
 follows:
 
  # Generated by NetworkManager
  domain providence.uwimona.edu.jm
  search providence.uwimona.edu.jm localdomain
  nameserver 172.18.196.1
 
 What should I be trying?? :)
 
 --
 Help kids everywhere map their world, at http://olpcMAP.net !
 
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel
 
 
 -- 
 Help kids everywhere map their world, at http://olpcMAP.net !
 
 
 

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] XS procurement recommendations (new hardware - Nosy Komba, Madagascar)

2012-04-18 Thread John Watlington

Just a quick reminder that we have XO-1.75 production units available for
people who want to work on XS on ARM.   Just fill out the form at:

   http://wiki.laptop.org/go/Contributors_program/Project_proposal_form

and list XS on ARM XO as the project.   If you have any problems getting
one this way, drop me a note.

Cheers,
wad

On Apr 17, 2012, at 1:52 PM, George Hunt wrote:

 Hi Kevin,
 
 As far as I can tell, the system programmers in Boston are very occupied with 
 getting 1.75 and 3.0 out the door.  I think there's interest in seeing an ARM 
 XS, but not immediately, with the resources available to them.
 
 There's a volunteer in San Francisco, who's email is rihoward, who has played 
 with a number of ARM boxes. He's already given me some really useful advice.
 
 I don't have the background to be very productive in getting the XS onto ARM, 
 but to me it seems strategic.
 
 I am spending a few hours every day, trying to learn about cross compiling 
 fc17 onto the ARM arch. 
 
 But I'm no expert.  Maybe we could join forces.  As I go along, one of my 
 goals it to document the steps at schoolserver.wordpress.com.
 
 George
 
 On Tue, Apr 17, 2012 at 12:28 PM, Kévin Raymond shai...@fedoraproject.org 
 wrote:
 On Tue, Apr 17, 2012 at 3:46 PM, Mitchell Seaton msea...@ekindling.org 
 wrote:
  I might sum up here from recent additions (replies) on this mailing list
  regarding current low-power server options (x86/32bit) for XS:
 
  1) FitPC2i - Dual Ethernet, fanless, poss heat-issue (heat sink),
   PhoenixBIOS (no Wake-on-LAN), 12W draw (poss ~5W idle)
  http://www.fit-pc.com/web/fit-pc/fit-pc2i-specifications/
 
  2) SolidLogic - (VIA C7, 25W draw) current sys in use not available for
  purchase, EU options in links below
  http://www.logicsupply.eu/systems/intel-atom/
  http://www.logicsupply.eu/systems/fanless/
  deployment sites: Jamaica
 
  3) MSi WindBox - '10 Atom D510, 4GB DDR2 - ~35W, drawback single Ethernet
  NIC (USB WiFi adapter)
  http://fr.msi.com/models/Wind%20Box%20DC500
  deployment sites: Nepal, Haiti, Philippines
 
  4) EPC-AT270 - Dual Ethernet (Wake on LAN), fanless, drawback '08 model Atom
  N270 (2GB DDR2) -
  http://www.avalue.com.tw/products/EPC-AT270.cfm
  deployment sites: PNG, Oceania?
 
  These solutions generally haven't changed in the last two years, in Q2, 2012
  now these are the current stable options for low-power x86?
 
  If that's the reality, I assume we can make a decision around these options,
  taking site specific considerations into account, and plan for possible
  involvement (replacement) down the line (1-2+ years) with any future server
  development for ARM/plug/etc and what suitable working model
  is devised around it.
 
  Regards,
  Mitch
 
 
 Hi there,
 
 I am new in this list but involved in OLPC-France.
 Could someone point me to the XS ARM port status?
 I don't know yet what software the XS is composed of, but since I have
 the Alex's dreamplug here in Paris I could try to move things forward.
 (Won't probably make it before May but costs nothing to read about it).
 
 I also have a FitPC(2?), but the power supply is lost somewhere in my bedroom…
 Yeah, it is way too hot IMHO
 
 --
 Kévin Raymond
 (shaiton)
 GPG-Key: A5BCB3A2
 
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Looking for new low power server hardware candidate

2012-04-10 Thread John Watlington

Why not an XO-1.75 ?

On Apr 10, 2012, at 8:04 PM, Martin Langhoff wrote:

 On Mon, Apr 2, 2012 at 2:03 AM, George Hunt georgejh...@gmail.com wrote:
 I left my fitpc2 and msi servers in the Philippines, hoping they would be
 pressed into service in a classroom situation.  So now I'm in the market for
 another toy.
 
 If you have time towork with us through some hitches, I'd recommend an
 ARM server. At this stage I'd say one of the Marvell/Globalscale
 Plug servers (dreamplug for example), or a trimslice.
 
 Either option will need a combination of the OS on internal SD/eMMC
 and the storage on an ext HDD (via USB probably).
 
 cheers,
 
 
 m
 -- 
  martin.langh...@gmail.com
  mar...@laptop.org -- Software Architect - OLPC
  - ask interesting questions
  - don't get distracted with shiny stuff  - working code first
  - http://wiki.laptop.org/go/User:Martinlanghoff
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel
 

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] XS on XO

2012-02-29 Thread John Watlington

How is this work going ?

On a somewhat related note, I was building a firewall/serve box from an XO-1.5
running os883, and ran into the problem that the stock kernel doesn't have
enough of the netfilter options enabled (specifically 
CONFIG_IP_NF_TARGET_REJECT)
to support either the firewall rules I wanted to use or the old XS firewall 
rules.

Daniel,
   How did you deal with this on the previous XS on XO work ?

Cheers,
wad

On Feb 23, 2012, at 4:52 PM, Martin Langhoff wrote:

 On Feb 23, 2012 10:27 AM, Daniel Drake d...@laptop.org wrote:
  Another option you may wish to explore is running CentOS6 with the
  OLPC kernel on the XO. (but I haven't really thought this through,
  might be missing something obvious)
 
 That sounds like a good approach to me. Given the python incompat, I'd try 
 this next.
 
 cheers,
 
 
 m 
 { Martin Langhoff - one laptop per child } 
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] A quick networking question

2012-02-29 Thread John Watlington

On Feb 29, 2012, at 10:05 AM, Samuel Greenfeld wrote:

 The example DHCP configuration linked likely should be updated to support 
 multiple MAC address ranges.
 
 In addition to the 00:17:C4 prefix mentioned in that script, newer XOs may 
 come with Wifi cards that have a 20:7C:8F prefix, and I'm looking at an XO 
 that has a 68:A3:C4 prefix.

The days of simple MAC filtering are probably over.As Samuel points out,
there are at least two sources of WLAN modules for XO-1.5/1.75 laptops, and
more might be introduced in the future.

There are well established protocols and services for authenticating access to a
network (802.1x).   Supposedly wpa_supplicant already supports some of them ?
It is time to start using one of them.

Cheers,
wad
___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] A quick networking question

2012-02-28 Thread John Watlington

On Feb 28, 2012, at 1:05 PM, Holt wrote:

 Clarif: port 80 is (unfort) forwarded thru the XS, for all laptops that 
 connect over Wifi.
 
 Traffic across all other ports (incl 443 = https) is thankfully blocked, 
 though I've no idea why/how unfortunately ;)

Sounds like your problem is squid.   Your firewall is probably blocking 
FORWARDS from non-XOs,
but routing all http traffic into squid.   You instead need to only route XO 
http traffic into squid.

What version school server software ?

Cheers,
wad

 On 2/28/2012 12:49 PM, Holt wrote:
 On 2/28/2012 12:29 PM, George Hunt wrote:
 In Haiti, Adam and I have been trying to get a school server online.  We're 
 finding that volunteers are going through the school server to the internet 
 with their laptops, and he wants to turn that off, at least for now.
 
 I've turned off /proc/net...ip_forward and verified that there is no 
 masquerade enabled in the iptables.
 
 But that's not enough!!  I wasn't sure that the vpn wasn't setting up a 
 gateway, so I had him turn off the vpn.  But still the school server was 
 routing to the 3G usb modem dongle even with the vpn pipe closed down.
 
 How does the school server act like a router?  It may be related to the ppp 
 connection and wdial configuration.  But I'm stumped.
 
 But I'm trying to bring myself up to speed quickly because he really wants 
 to get it turned off.
 
 Any ideas on what to try next?  I'm afraid the solution is going to be to 
 pull out the 3g dongle.
 
 Interestingly the XS(*) creates an open path for any random non-XO laptop to 
 access the web, but seems to block non-web traffic like ssh and IMAP.
 
 In any case, even if it's just forwarding port 80 and 443 (?) we just cannot 
 afford to become a free ISP here in semi-rural Haiti, given so many visitors 
 to our school especially.
 
(*) XS as set up by Tony Anderson early autumn 2011, and currently 
 maintained by George Hunt  I.
 
 --
 Help kids everywhere map their world, at http://olpcMAP.net !
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Who wrote http://wiki.laptop.org/go/XS_Install_Server?

2012-01-30 Thread John Watlington

On Jan 30, 2012, at 12:40 PM, George Hunt wrote:

 Hi everyone,
 
 I'm new to this list. 
 
 I met Tony Anderson in Haiti, and again at the San Francisco OLPC Summit in 
 late 2011. He prevailed upon me to spend some time trying to figure out how 
 to rebase XS on a more recent Fedora Core.
 
 I just recently used pungi to gather together FC16 rpms into an iso using the 
 kickstart file that came with XS-0.6.  What was 550MB in FC9 has grown to 
 775MB in FC16.  My initial goal was to create a CD image, that can be 
 installed in the traditional way. But going forward, even if we can pare 
 things down and fit in 690MB this time, perhaps we need a better strategy.
 
 I came across comments in the XS wiki  to the effect that USB installs don't 
 always work, suggesting that a CD installation was more reliable.  
 
 Then I discovered this page (which I had missed before) which displayed a 
 level of knowledge, and perspective, which I would like to tap.
 
 Can anyone own up to being the author, or point to him/her?

Here is the relevant email thread, including the author's email and
his hindsight:

Begin forwarded message:

 From: Tim Moody timmo...@sympatico.ca
 Date: April 29, 2011 9:15:42 AM EDT
 To: Sridhar Dhanapalan srid...@laptop.org.au
 Cc: server-devel@lists.laptop.org
 Subject: Re: [Server-devel] XS Install Server
 
 Looks like I could benefit from your kickstart file(s).  I didn't install 
 from USB because a) my machines are old and don't boot from USB and b) I 
 thought a server would scale better for large implementations, but clearly 
 USB is much more portable if your model is to install in the field.  In any 
 event, both of our approaches drive off the kickstart file.  I'd be happy to 
 lend a hand if there's something you can parcel out.  How do you validate 
 the XS install?  I don't actually have an XO, so I run SOAS in VirtualBox, 
 which is not a very thorough test.
 
 --
 From: Sridhar Dhanapalan srid...@laptop.org.au
 Sent: Friday, April 29, 2011 1:57 AM
 To: Tim Moody timmo...@sympatico.ca
 Cc: server-devel@lists.laptop.org
 Subject: Re: [Server-devel] XS Install Server
 
 On 29 April 2011 15:43, Sridhar Dhanapalan srid...@laptop.org.au wrote:
 On 9 April 2011 03:16, Tim Moody timmo...@sympatico.ca wrote:
 I have been experimenting for quite awhile with an install server based 
 on
 cobbler and have finally written it up here
 http://wiki.laptop.org/go/XS_Install_Server.
 
 If you have more than a few servers to install and want to automate the
 process
 
 or you want to add some library content automatically, you may find this 
 of
 interest.
 
 Interesting work!
 
 The approach we have taken with our XS-AU is to configure the
 installation via kickstart files and install from USB.
 
 http://dev.laptop.org.au/projects/xs-au/wiki
 
 I should also add that I think we have already achieved some of your
 goals. Looking at your Future list at
 http://wiki.laptop.org/go/XS_Install_Server#Future:
 
 * Create two separate distros, one for FC and the other for XS and
 declare them in the kickstart file to separate the XS packages from
 Fedora
 
 The XS-AU is a modular set of packages on top of Fedora 11:
 http://dev.laptop.org.au/projects/xs-au/wiki/Install_on_an_existing_Fedora_installation
 
 * Upgrade Fedora to latest version
 
 Once we release our version 1.0 (based on Fedora 11), we can probably
 move to Fedora 13 with minimal effort. Fedora 14 will be trickier, as
 it has been upgraded to Python 2.7.
 
 * Create separate kickstart files for full install,
 networking/gateway only install, services only install
 
 Our installation is completely kickstart-driven, so we are most of the
 way there.
 
 We have other improvements in the pipeline as well.
 
 We'd be happy to work with anyone interested in getting involved.
 
 Sridhar
 
___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Unable to Browse the Internet from XO

2012-01-29 Thread John Watlington

On Jan 27, 2012, at 1:45 PM, HALL,Brian C wrote:

 Good Day All,
 
 I am currently using the 1.5 XO at a couple of schools. I am  able to connect 
 to the school server and hence access moodle instance on the server itself. 
 However i am unable to browse the internet from the XO.

Can you provide a little more information about your installation ?

 I have connected the ETH1 to the ISP switch and still no success.

What IP address gets assigned to the ETH1 interface on the school server ?
(Type ifconfig on the server to get this info)

Can you access the internet from the school server itself ?

What happens when you try to access the internet from the laptop ?
Does DNS work ?  (what happens when you type: dig www.apple.com ?)
Can you ping outside the school server internal network ?

Cheers,
wad

On Jan 27, 2012, at 1:45 PM, HALL,Brian C wrote:

 Good Day All,
 
 I am currently using the 1.5 XO at a couple of schools. I am  able to connect 
 to the school server and hence access moodle  instance on the server itself. 
 However i am unable to browse the internet from the XO.
 
 
 
 
 
 I have connected the ETH1 to the ISP switch and still no success.
 
 
 Any help would be greatly appreciated.
 
 
 Regards,
 Brian Hall
 ___
 Devel mailing list
 de...@lists.laptop.org
 http://lists.laptop.org/listinfo/devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Help buying server for XS

2011-03-28 Thread John Watlington

On Mar 27, 2011, at 11:28 PM, Guillermo Narvaez wrote:

 Reuben,
 
 The limitations of this equipment are 60 client per AP, we are using 50 or 55 
 clients per equipment. This is a dedicated equipment so it need an external 
 controller.
 
 With the frequency issue, the APs of the same school are set in 1, 5, 9 and 
 11 channels.

We can't recommend this configuration.   There is spectrum overlap between
channel 1 and 5, 5 and 9, and 9 and 11.   Transmissions in channel 1 will 
interfere
with those in channel 5.   Not completely, thanks to the design of the 802.11 
wireless
layer, but enough to significantly degrade the bandwidth available if they are 
physically
coexistant.

We instead recommend working with channels 1, 6, and 11, which theoretically
have no overlap with one another.

Given the consumer nature of the RF filters in WiFi units, you are best
using only channels 1 and 11 if using two frequencies, to avoid cross channel
interference.When testing in a dense environment (a classroom), I've seen 
WiFi
units tuned to channel 1 receive packets transmitted on channel 6.

Regards,
wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] callhome script

2011-03-08 Thread John Watlington

callhome was a quick hack.
I would suggest doing it right instead of recycling it.

Cheers,
wad

On Mar 8, 2011, at 4:21 AM, Sameer Verma wrote:

 Found the callhome script on the XS (0.6). Has anyone used it in the
 field? Is it documented/recommended? I'm looking at it as an
 alternative to setting up a openvpn tunnel from an XS to a public IP
 someplace.
 
 cheers,
 Sameer
 -- 
 Dr. Sameer Verma, Ph.D.
 Associate Professor, Information Systems
 Director, Campus Business Solutions
 San Francisco State University
 http://verma.sfsu.edu/
 http://opensource.sfsu.edu/
 http://cbs.sfsu.edu/
 http://is.sfsu.edu/
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] IP Address Pools for XOs, known clients, and unknown clients on XS 0.6

2011-01-12 Thread John Watlington

The best iptables hack like this I've seen routed extraneous
connections through a transparent web proxy which flipped
all images (swapped left and right).

Cheers,
wad

On Jan 12, 2011, at 11:46 AM, Jerry Vonau wrote:

 On Wed, 2011-01-12 at 10:03 -0600, Anna wrote:
 I like to leave the AP open on my test XS 0.6 at home, but ran into an
 issue with that yesterday.  I noticed the lights on my router blinking
 like crazy, so I did a live tail on the squid access log to see what
 was going on.
 
 tail -f /var/log/squid/access.log
 
 snip
 And because I'm ticked off, and inspired by
 http://www.ex-parrot.com/pete/upside-down-ternet.html, it's time for
 some fun with iptables.  In /etc/sysconfig/olpc-scripts/iptables-xs.in
 I add a couple of lines like so:
 
 So I'm not the only one who likes fun with iptables, wish I could see
 the expression on their face when I tried something like that. 
 
 *nat
 :PREROUTING ACCEPT [0:0]
 :POSTROUTING ACCEPT [0:0]
 :OUTPUT ACCEPT [0:0]
 -A PREROUTING -s 172.18.124.0/24 -p tcp --dport 80 -j DNAT --to
 205.196.209.62
 @@SQUID@@
 -A POSTROUTING -o @@WAN@@ -j MASQUERADE
 COMMIT
 *filter
 :INPUT ACCEPT [0:0]
 :FORWARD ACCEPT [0:0]
 :OUTPUT ACCEPT [0:0]
 -A FORWARD -s 172.18.124.0/24 -p tcp --dport 443 -j DROP
 
 This should take care of the rest of the outgoing connections..
 change to: 
 -A FORWARD -s 172.18.124.0/24 -p tcp ! --dport 80 -j DROP
 
 add:
 -A FORWARD -s 172.18.124.0/24 -j DROP
 
 COMMIT
 
 Restart dhcpd and iptables:
 service dhcpd restart
 service iptables restart
 
 Now all unknown clients will have http traffic redirected to
 http://kittenwar.com and their https traffic is dropped.
 
 Obviously this isn't a deterrent to someone who can use an ssh proxy
 for browsing, and it doesn't block traffic on other ports or
 protocols, but most of my neighbors aren't of the networking savvy
 sort (particularly the grotesque rednecks) and will likely conclude
 this darn internet ain't workin' no more.  If I lived near MIT, this
 would not be an acceptable solution.  But I'm not terribly concerned
 many folks around here know much about packet sniffing or MAC
 spoofing.
 
 
 His machine might be owned/spam-bot... Try the trivial change above.
 
 When guests come over and want to look at something other than
 pictures of kittens, all I have to do is add the MAC to the list of
 known clients, restart dhcpd, and tell them to renew their IP.
 
 At the very least, now I know how to keep XOs and non-XO clients on
 different IP ranges.
 
 Anna Schoolfield
 Birmingham
 
 Jerry
 
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] XS on XO Setup as Contingency for Main Power Outage

2010-12-15 Thread John Watlington

On Dec 15, 2010, at 7:09 PM, Anna wrote:

 Sorry this is so long, but I thought I'd post this in case someone else 
 needed something similar.  Also, any suggestions to make this more efficient 
 are more than welcome.
 
 How I set up an emergency use XS on an XO-1
 
 Since I run a public XS on a big old Dell in my house, I needed a solid 
 contingency plan should the power go out.  While I have my XS on a UPS, it's 
 small and cheap and only keeps it up about half an hour.  I do have a power 
 inverter that we hook up to a car battery, which keeps the DSL modem, and 
 thus our internet, up for many hours so we can use our netbooks.  Charging an 
 XO shouldn't add that much more drain.  So to keep my users happy and 
 chatting while I surf the web in the dark, I set up the XS on an XO-1.  This 
 will also be useful in the summer should my A/C unit go out when it's 100 
 degrees and I don't want my regular equipment to overheat.
 
 I followed the instructions here to set up an 8 GB SD card:  
 http://wiki.laptop.org/go/XS-on-XO
 It takes forever for the first boot, of course, while moodle initializes. 

I would recommend that you use a class 6 or class 10 full size SD card for this 
purpose.
One of the Sandisk Extreme III cards, for example.   The extra cost is worth it 
for the server.
There is a huge difference in card performance, especially for small file 
writes, and
the new larger sized (8+GB) microSD cards tend to be especially bad.

Great write-up, by the way.

Thanks!
wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Firstboot modules

2010-08-27 Thread John Watlington

On Aug 26, 2010, at 8:04 PM, Martin Langhoff wrote:

 StPepper,
 
 forget about anaconda :-)  I want to _reduce_ as much as reasonable
 what we do in anaconda/kickstart.
 
 Let's focus on this:
 
 - some stuff in current kickstart %post won't be needed for xs-0.7
 
 - some stuff in current kickstart can be sanely moved to a firstboot script
 
 - other stuff (like network config) should be handled somewhere else
 -- and here I strongly prefer a combination of Moodle admin UI taking
 the values and writing them to a pre-arranged file. Scripts running as
 root can then read those files, check for sanity, and apply the
 changes. We can use incrond to trigger those.

Moodle takes over yet another part of the world...

Martin,
   Over a year ago I thought we agreed on grabbing those parameters
at first boot from a file, either provided through a USB key or placed on
the disk during initialization.   I guess that plan fell through ?

Cheers,
wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] wifi setup

2010-08-03 Thread John Watlington

I was under the impression that this required 802.11F
capable APs to work properly.
But I'm no expert.  Anyone better informed care to educate
me ?

Cheers,
wad

On Aug 3, 2010, at 11:48 PM, James Cameron wrote:

 Different OLPC OS builds present access points with same essid in
 different ways.  There was a problem with multiple access point
 presentation fixed somewhere along the way in development, but I'd have
 to go looking for it.  Which build are you working with?
 
 (Tabitha mentioned 8.2.1 on Monday ... and I think that build is
 affected by the problem).
 
 You could also capture the output of command
 
   sudo iwlist eth0 scan  scan.txt
 
 and check the file carefully for anything strange in the essid that
 might make them unique.
 
 -- 
 James Cameron
 http://quozl.linux.org.au/
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Upgrading XO build 656 failed

2010-04-10 Thread John Watlington

This is the wrong list for such a question,
it is for questions about the school server.
Your question should be posted to de...@lists.laptop.org,
where the right people are more likely to see it.

If you aren't a member of either list, the message is
held for moderation (and our moderators are slow).
You can add yourself to the lists at:
http://lists.laptop.org/listinfo/

We need more information before we can give a good answer:

- Does your laptop have OLPC security enabled ?
  (where did you get it ? would answer this question...)
  Did you ever install a developer key and disable-security ?

- How did you upgrade it ?
  (olpc-upgrade doesn't really work in build 656)
  Did you upgrade it from Open Firmware ?

Regards,
wad

On Apr 10, 2010, at 2:32 AM, ganesh gajre wrote:

 Hello,
 
 I am trying to upgrade some XO which have build 656. When i tried to
 upgrade it with build 703.
 
 It's shows me error as Can't find NAND image.
 
 I upgrade firmware also, which is Q2d16.rom
 
 Can someone help me to solve this problem.
 
 -- 
 Ganesh (Dragger)
 Be a FOSSERS, use GNU/Linux
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] eth0 Issue on Acer Aspire One

2010-04-10 Thread John Watlington

From the intertubes, it looks like the AAO has an Atheros 5K chipset.
Lots of problems reported with the drivers, but people have gotten it
to work with Fedora 9.  E.g.:
http://www.mail-archive.com/ath5k-us...@lists.ath5k.org/msg00131.html

Good luck,
wad

On Apr 10, 2010, at 10:02 PM, Andra DuPont wrote:

 I am considering using an Acer Aspire One as an XS. I have installed XS 0.6, 
 but I'm having trouble with the wired ethernet port. It is not recognized as 
 eth0. If I plug in a USB to Ethernet converter it is recognized as eth0 but 
 there is no eth1.
 
 Is this an issue of drivers for the ethernet port on the Aspire?
 
 With the USB converter unplugged,
 
 # ethtool -i eth0 says No such device.
 
 Any thoughts on how to trouble shoot this. I am considering getting a second 
 USB to Ethernet converter and using both to get eth0 and eth1.
 
 Andy
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Web Caching for Areas with Intermittent

2010-04-09 Thread John Watlington

On Apr 9, 2010, at 8:38 PM, John Watlington wrote:

 Last time I looked into this, apache's proxy was solely a reverse proxy,
 and useless for our purposes.   That is no longer true, but the associated
 mod_cache module still appears to be experimental.
 In any case, squid currently provides the forward proxy/caching service
 on the XS.
I forgot the key issuen here: it is difficult to use apache as a transparent
forward proxy...

 The issue is that most proxies cease functioning if they aren't
 connected to the internet.   They only cache cacheable pages, etc.
 
 Search back through the mailing list archives for Andra original
 posts...
 wad
 
 On Apr 9, 2010, at 10:25 AM, Tim Moody wrote:
 
 What are the main advantages of wwwoffle over apache's built in proxy and 
 cache modules?
 
 Message: 3
 Date: Wed, 7 Apr 2010 23:46:52 -0400
 From: Andra DuPont andradup...@gmail.com
 Subject: [Server-devel] Web Caching for Areas with Intermittent
 Internet - WWWOFFLE installation Guide on 0.6 XS Server
 To: XS Devel server-devel@lists.laptop.org
 Message-ID: 8959cbcb-d54b-4fb6-996a-e4a403dc9...@gmail.com
 Content-Type: text/plain; charset=us-ascii
 
 Squid apparently is not well suited to serve up cached pages when no 
 internet access is
 available. It was recommended that I use WWWOFFLE for this situation. Here 
 is the
 process I used to get it installed and running on my 0.6 XS server:
 
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel
 
 

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] ssh connection to XS Server

2010-03-30 Thread John Watlington

On Mar 30, 2010, at 2:10 AM, David Mul wrote:

 Hi,
 
 my name is David Mul and I'm working with David Leeming on the OLPC project 
 in Papua New Guinea. I was trying to login to the XS server using terminal 
 activity on the XO through ssh but it kept giving me this message Permission 
 denied (publickey).
 Is there a solution to this problem.

As what user on the XS ?

Quick answer might be to edit /etc/ssh/sshd_config to change to 
PasswordAuthentication yes

Cheers,
wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Web Caching Issues

2010-03-21 Thread John Watlington

You don't really want to disable or move the idmgr.
All the laptop software builds will try to contact it in
order to register themselves.   Moving it would require
doing a special software build for all the XO laptops in your
school, and providing a separate update path

wad

On Mar 21, 2010, at 2:19 PM, Jerry Vonau wrote:

 On Sun, 2010-03-21 at 13:56 -0400, Andra DuPont wrote:
 Great... thanks.
 
 I have now been able to get a version of wwwoffle installed, but
 when I try to start it, it fails because it is not able to server socket
 0.0.0.0 port 8080 because it is already in use. I have not started
 squid, so I'm guessing that the XS is using it. Andy ideas on how to
 free it up for wwwoffle?
 
 Think that is idmgr, you'll need to disable that, if you do run wwwoffle
 on port 8080. you'll need to change the iptables redirect to match port
 8080. You could configure wwwoffle to use port 3128 instead, then you
 should have a drop-in replacement for squid.
 
 Jerry
 
 
 
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Web Caching Issues

2010-03-16 Thread John Watlington

When you turn on offline mode, you are just handling
the HTTP requests.   There are other network operations
which have to happen before hand which probably aren't
being offlined properly.

What happens to DNS in the offline scenario ?  If your clients
are all properly pointed at the school server's DNS server,
and the entries aren't timing out, it might resolve properly.
Otherwise, the HTTP request will never be made.

Once you switch to offline mode, what does squid's access
log indicate ?   error log ?   You can turn on logging on the
DNS server and see if name resolution is being requested.

Cheers,
wad

On Mar 16, 2010, at 10:38 PM, Andra DuPont wrote:

 I have reconfigured my server so that my internet connection goes straight 
 from my DSL modem to eth0 on the XS. This did not solve the problem (web 
 chaching not working), but it eliminated Windows 7 Firewall as the cause. I 
 have several computer connected both wired and wirelessly to the AP on eth1 
 of the XS. They can access both the schoolserver and the internet (so long as 
 the internet is active on eth0 of XS). When I disconnect the internet 
 service, the AP computers can still get to the schoolserver, but not to the 
 external web pages that should have been cached from the internet.
 
 Internet service at the remote village in Kenya where this server will be 
 installed is poor and intermittent. It is therefore critical to be able to 
 cache whatever content does come in. I am also considering some modifications 
 to the Squid.conf file to accommodate turning the offline_mode on and off 
 depending on the status of the internet connection. See this article for the 
 concept and proposed solution.

See what article ?

 I have tried turning offline_mode on after caching a bunch of web pages to 
 see what would happen. When I disconnected the internet service to eth0 on 
 the XS, external web service stopped. In other words, the server would not 
 serve up the cached material, or at least that was my conclusion.
 
 Does anyone have a simple diagnostic approach to this problem. I am new to 
 linux, OLPC and XS, so it's like drinking from a fire hose right now. Can 
 someone point me in the right direction.
 
 On Mar 16, 2010, at 2:51 PM, Andra DuPont wrote:
 
 I have set up an XS server at my home to become familiar with it so I can 
 help an installation in Asilong, Kenya. There server has never been used as 
 no one there understands it.
 
 My server is running the latest XS (0.6) and in order to simulate the 
 situation in Asilong, the server is connected to the internet on eth0 from a 
 PC running Windows 7. The PC is connected wirelessly to my home router, and 
 that connection is bridged on the PC to its ethernet port which is then 
 connected to the XS-eth0. The XS-eth1 is connected to a wireless AP and I'm 
 testing performance of the server by connecting an iMac to the AP. The iMac 
 can log into the Schoolserver, and can also access the internet. So far so 
 good.
 
 I have enabled squid on my server, but web caching does not seem to be 
 working. I'm wondering if feeding my internet connection to the server from 
 a bridged connection on my Windows 7 computer with Windows Firewall running 
 is interfering with some of the dns information on the cache.
 
 Revisited pages don't load faster (if I clear the cache on the connected 
 iMac), and if I disconnect the eth0 internet feed, the server is unable to 
 retrieve anything external, even material that should be cached.
 
 I am working now on setting up the forwarders correctly in the 
 named-xs.conf.in file, and also playing with the offline_mode on setting 
 in the squid-xs.conf file to see what affect it has.
 
 Any thoughts from the team?
 
 Andy DuPont
 
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Web Caching Issues

2010-03-16 Thread John Watlington

On Mar 16, 2010, at 11:30 PM, Andra DuPont wrote:

 With the internet disconnected, my web browser just says it can't display 
 anything
 because it is not connected to the internet. When I plug the DSL modem back 
 into
 eth0 on the XS, all is fine.
 
 How do I turn on DNS server logging?

You increment the trace level using the remote name daemon controller:
Something like:
sudo rndc trace 5

 Andy
 
 On Mar 16, 2010, at 11:06 PM, John Watlington wrote:
 
 
 When you turn on offline mode, you are just handling
 the HTTP requests.   There are other network operations
 which have to happen before hand which probably aren't
 being offlined properly.
 
 What happens to DNS in the offline scenario ?  If your clients
 are all properly pointed at the school server's DNS server,
 and the entries aren't timing out, it might resolve properly.
 Otherwise, the HTTP request will never be made.
 
 Once you switch to offline mode, what does squid's access
 log indicate ?   error log ?   You can turn on logging on the
 DNS server and see if name resolution is being requested.
 
 Cheers,
 wad
 
 On Mar 16, 2010, at 10:38 PM, Andra DuPont wrote:
 
 I have reconfigured my server so that my internet connection goes straight 
 from my DSL modem to eth0 on the XS. This did not solve the problem (web 
 chaching not working), but it eliminated Windows 7 Firewall as the cause. I 
 have several computer connected both wired and wirelessly to the AP on eth1 
 of the XS. They can access both the schoolserver and the internet (so long 
 as the internet is active on eth0 of XS). When I disconnect the internet 
 service, the AP computers can still get to the schoolserver, but not to the 
 external web pages that should have been cached from the internet.
 
 Internet service at the remote village in Kenya where this server will be 
 installed is poor and intermittent. It is therefore critical to be able to 
 cache whatever content does come in. I am also considering some 
 modifications to the Squid.conf file to accommodate turning the 
 offline_mode on and off depending on the status of the internet connection. 
 See this article for the concept and proposed solution.
 
 See what article ?
 
 I have tried turning offline_mode on after caching a bunch of web pages 
 to see what would happen. When I disconnected the internet service to eth0 
 on the XS, external web service stopped. In other words, the server would 
 not serve up the cached material, or at least that was my conclusion.
 
 Does anyone have a simple diagnostic approach to this problem. I am new to 
 linux, OLPC and XS, so it's like drinking from a fire hose right now. Can 
 someone point me in the right direction.
 
 On Mar 16, 2010, at 2:51 PM, Andra DuPont wrote:
 
 I have set up an XS server at my home to become familiar with it so I can 
 help an installation in Asilong, Kenya. There server has never been used 
 as no one there understands it.
 
 My server is running the latest XS (0.6) and in order to simulate the 
 situation in Asilong, the server is connected to the internet on eth0 from 
 a PC running Windows 7. The PC is connected wirelessly to my home router, 
 and that connection is bridged on the PC to its ethernet port which is 
 then connected to the XS-eth0. The XS-eth1 is connected to a wireless AP 
 and I'm testing performance of the server by connecting an iMac to the AP. 
 The iMac can log into the Schoolserver, and can also access the internet. 
 So far so good.
 
 I have enabled squid on my server, but web caching does not seem to be 
 working. I'm wondering if feeding my internet connection to the server 
 from a bridged connection on my Windows 7 computer with Windows Firewall 
 running is interfering with some of the dns information on the cache.
 
 Revisited pages don't load faster (if I clear the cache on the connected 
 iMac), and if I disconnect the eth0 internet feed, the server is unable to 
 retrieve anything external, even material that should be cached.
 
 I am working now on setting up the forwarders correctly in the 
 named-xs.conf.in file, and also playing with the offline_mode on setting 
 in the squid-xs.conf file to see what affect it has.
 
 Any thoughts from the team?
 
 Andy DuPont
 
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel
 
 
 

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Collaboration server for existing network

2010-03-15 Thread John Watlington

On Mar 15, 2010, at 1:18 PM, Martin Langhoff wrote:

 On Mon, Mar 15, 2010 at 12:09 AM, John Watlington w...@laptop.org  
 wrote:
 And indeed, the XS services are intended for such reuse.   I'm not  
 sure
 how many patches to the stock ejabberd are still needed...

 It's not so much the patches (you can grab the ejabberd-xs rpm from
 our repo), but the integration with other aspects.

 ejabberd is integrated with moodle which is integrated with
 registration. The basics are registration + moodle + ejabberd. Less
 than that and you really don't have the goods.

I expected that you would have to grab other parts as well.
But most of the underlying tools are relatively distro-independent.

 (Yes, you can get just get our patched ejabberd, but configuring it is
 messy, and it'll only work nicely with a small school, say less than
 50~60 users.)

Yep.  Configuring everything to interoperate is the special sauce!

Cheers,
wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Collaboration server for existing network

2010-03-14 Thread John Watlington

On Mar 15, 2010, at 12:06 AM, James Cameron wrote:

 I don't know XS very well, but if ejabberd is all you need why not  
 take
 the ejabberd configuration from XS sources and deploy that on an
 otherwise vanilla instance?

And indeed, the XS services are intended for such reuse.   I'm not sure
how many patches to the stock ejabberd are still needed...

The main hook needed in the existing infrastructure is a DNS entry
for the machine running the school server services.   Those services
can either be easily grafted on top of a Fedora server using the school
server packages, or supported on another Linux distro with some hand
tweaking.

wad
___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] XS on Mac Mini

2010-03-02 Thread John Watlington

It would be interesting to hear what the remaining hurdles are.

The Kirkwood ARM processor in the guruplug should be well
supported by the current Fedora release --- this may be mostly
a matter of moving to an F12 base.

My main concern with the guruplug would be the external USB
disks (overheating or disappearing).   The same hardware should
be available in a NAS (networked attached storage unit), wrapped
around a hard drive in an enclosure --- the Marvell ARM is a popular
choice.

We have good relations with Marvell, and have been encouraging
them to get support for their ARM processors upstream for
several years now.

wad

On Mar 2, 2010, at 12:27 PM, Sameer Verma wrote:

 On Tue, Mar 2, 2010 at 6:09 AM, Caroline Meeks
 carol...@solutiongrove.com wrote:
 I was talking to Timothy Falconer who is planning deployments in  
 Haiti
 and Patrick Giagnocavo founder of Zill.net our sysadmin and  
 hosting company
 about this.
 Patrick suggests:  http://www.globalscaletechnologies.com/c-4- 
 guruplugs.aspx
 the specs on the Plus version that is $129
 2x Gigabit ethernet plus Wifi
 But I don't think any one has tried it yet.

 These are ARM based, right? Robert Howard @OLPCSF (on the list) has
 been tackling this a bit, but there are hurdles with running XS on
 ARM...

 Sameer


 On Tue, Mar 2, 2010 at 8:38 AM, Martin Langhoff  
 martin.langh...@gmail.com
 wrote:

 On Tue, Mar 2, 2010 at 1:19 AM, David Leeming
 da...@leeming-consulting.com wrote:
 Sorry if this has been asked before. The Mac Mini was tested by  
 someone
 here
 with a watt meter and ran between 16 and 30 watts max. Very  
 useful for
 solar
 powered locations.

 Agreed with Peter -- it'll just work, but other boxes may be  
 cheaper,
 and most importantly, may be better suited to hot  humid
 environments.

 Whatever you use, let us know how it goes!


 m
 --
  martin.langh...@gmail.com
  mar...@laptop.org -- School Server Architect
  - ask interesting questions
  - don't get distracted with shiny stuff  - working code first
  - http://wiki.laptop.org/go/User:Martinlanghoff
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel



 --
 Caroline Meeks
 Solution Grove
 carol...@solutiongrove.com

 617-500-3488 - Office
 505-213-3268 - Fax

 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel


___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] XS on Mac Mini

2010-03-02 Thread John Watlington

On Mar 2, 2010, at 9:48 PM, Martin Langhoff wrote:

 On Tue, Mar 2, 2010 at 8:40 PM, John Watlington w...@laptop.org  
 wrote:
 It would be interesting to hear what the remaining hurdles are.

 Not many. There are good bootable images for F11/F12 on ARM. Current
 XS is on F9, for which I haven't found one.

 So it would be

  - port / test the XS packages on F11/F12
  - rebuild ejabberd-xs -- the only binary package we have

For some reason I find it humorous that we actually have
to rebuild erlang, not ejabberd-xs...   But erlang has already
been rebuilt, as it is standard with Fedora.

  - test the whole thing

Ay, the rub is in the details

 cheers,


 m
 -- 
  martin.langh...@gmail.com
  mar...@laptop.org -- School Server Architect
  - ask interesting questions
  - don't get distracted with shiny stuff  - working code first
  - http://wiki.laptop.org/go/User:Martinlanghoff

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Error message - etc/init.d/callhome restart

2010-02-23 Thread John Watlington

Remove the cron script for callhome, located in /etc/cron.d.
It isn't needed for your closely monitored setup, and if needed
should be revamped.

Cheers,
wad

On Feb 23, 2010, at 4:32 PM, Anna wrote:

 XS 0.6 has been sending me a ton of these emails.  Is there a fix  
 for this?

 -- Forwarded message --
 From: root usern...@gmail.com
 Date: Tue, Feb 23, 2010 at 12:00 PM
 Subject: Cron r...@schoolserver /etc/init.d/callhome restart
 To: root


 /etc/init.d/callhome: line 14: .: /etc/sysconfig/callhome: is a  
 directory
 /etc/init.d/callhome: line 17: [: =: unary operator expected
 Shutting down callhome: [FAILED]
 Starting callhome: [  OK  ]

 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] How to setup wireless mesh network for School server

2010-01-27 Thread John Watlington

School server installations using active antennas are
currently non-existant.

As the school server really wants to be a gateway, you will need
two network interfaces on your laptop.   You will need to connect
one of those to a WiFi AP for the XO(s) to connect to.

Theoretically, you should be able to turn your laptop's wireless
interface into that WiFi AP, but the appropriate software for
doing so is not part of the school server installation.

I suggest disabling the WiFi on the laptop and using two USB dongles.

Cheers,
wad

On Jan 28, 2010, at 12:09 PM, ganesh gajre wrote:

 Hello All,

 I had setup a schoolserver at my place using Lenovo laptop. All the  
 other stuff like backup-XO, moodle is configured. Now my problem  
 is, i am unable to setup a wireless mesh for communicating between  
 XS and XO.

 I don't have active antena, is it possible to create mesh using  
 wireless point of a laptop?   I had setup my WAN on wireless point,  
 but my requirement is to use USB internet dongle for WAN.

 Can anyone suggest me how to do  it?

 -- 
 Ganesh (Dragger)
 Be a FOSSERS, use GNU/Linux
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Backup of user home

2009-09-14 Thread John Watlington

You are correct.   Since your backup script by definition needs to  
read ALL data,
regardless of ownership or permissions, you will have to run it with  
superuser
priviledges.

Why don't you consider using the rsync server, instead of invoking it  
through a user ?

John


On Sep 14, 2009, at 3:26 PM, Rodolfo D. wrote:

 Hello:

 I'm working on a backup and restore feature for our schoolservers,  
 and I got stuck on home directories of laptops

 The backup works like this.. based on the backup script provided by  
 dsd, and also based on our specific features.. I placed all  
 important data in a directory /library/backup (wich can later be  
 tar, ziped, and in our case rsync-ed), and its being done by a cron  
 job

 On a centralized backup server.. we have a script that PULLS the / 
 library/backup of each server, so main configs are being saved  
 without much hassle.. But when it comes to user directories, it  
 lack permissions, because the /library/users/SN directory has no  
 read permissions for others

 how would you recomend that we do this?

 My first thought was to simply just add recursive read permissions  
 to the user folder.. but that doesn't take security in mind..  
 perhaps there's another way

 for now our pull works like this:

 r...@backupserver ~ $ rsync u...@schoolserver:/library/backup/ / 
 backup/schoolserver/backup/
 r...@backupserver ~ $ rsync u...@schoolserver:/library/users/ / 
 backup/schoolserver/users/

 r...@backupserver ~ $ rsync us...@schoolserver2:/library/backup/ / 
 backup/schoolserver/backup/
 r...@backupserver ~ $ rsync us...@schoolserver2:/library/users/ / 
 backup/schoolserver/users/

 Doing a push as a cron job from the server was a second idea, but  
 the backup server does other things so security in the backup  
 server is very important

 Any ideas?

 cheers..

 -- 
 Rodolfo

 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] /library partition

2009-07-18 Thread John Watlington

On Jul 16, 2009, at 4:40 PM, Sameer Verma wrote:

 Any documentation on what lives in /library and how to make it useful?

The idea behind /library was that the school server needed someplace to
place large collections of content and user data.

While some people have talked about backing up /library, my thinking
was almost the opposite: while other partitions would move to an SSD
for reliability,  /library would remain on a hard disk for a good $/ 
GB ratio.

Cost wise, a 4 GB SSD should be about $10.

Cheers,
wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Filtering and authentication

2009-04-27 Thread John Watlington

On Apr 27, 2009, at 11:18 AM, Reuben K. Caron wrote:

 Anna wrote:
 On Sun, Apr 26, 2009 at 10:51 AM, Reuben K. Caron  
 reu...@laptop.org wrote:

 As far as limiting the internet connection to authorized XOs,  
 that's an
 issue we're probably going to run into at some point once we  
 broaden the XS
 deployment.  So far at the pilot school, the staff members connect  
 to the
 internet with their personal laptops and iPhones, but I haven't  
 really heard
 any complaints of abuse yet.

 If your deployment is relatively small, it should be easy enough  
 to add the
 hardware addresses of the trusted XOs to dhcpd.conf and disallow  
 unknown
 machines (or play pranks on them as suggested at
 http://www.ex-parrot.com/~pete/upside-down-ternet.html).

 Anna Schoolfield
 Birmingham

 While not all encompassing you could also attempt to drop dhcp  
 requests
 that do not come from 00:17:c4 using something similar to:

 http://ubuntuforums.org/showthread.php?p=4191756

Please do not take this approach.   It sounds quick, easy, and  
foolproof,
but will lead to problems in the future.(I almost suggested it,  
but decided
the cons outweighted the pros.)

For example, what if you get an XO-1.5 in the mix ?  It won't work,  
and will
be difficult to debug.   You also disallow other laptops (teachers,  
etc.) from
being in the network...

wad



___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] mkusbinstall link broken?

2009-03-20 Thread John Watlington

The OLPC web-based git interface was changed, due to security
holes in the previous one.   Unfortunately, links into it still haven't
been changed.

On Mar 20, 2009, at 1:17 AM, Sameer Verma wrote:

 http://dev.laptop.org/git?p=projects/xs-livecd;a=blob;f=util/ 
 mkusbinstall;hb=HEAD

Git it in raw form at:
http://dev.laptop.org/git/projects/xs-livecd/plain/util/mkusbinstall
or formatted at:
http://dev.laptop.org/git/projects/xs-livecd/tree/util/mkusbinstall

 on http://wiki.laptop.org/go/ 
 XS_Installing_Software#Optional:_Using_a_USB_key. 
 2Fdisk_for_installation
 leads to a blank page.

Cheers,
wad

 Suggestions?

 Sameer
 -- 
 Dr. Sameer Verma, Ph.D.
 Associate Professor of Information Systems
 San Francisco State University
 San Francisco CA 94132 USA
 http://verma.sfsu.edu/
 http://opensource.sfsu.edu/
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] thoughts on registering SoaS with an XS and moodle integration

2009-03-12 Thread John Watlington

On Mar 12, 2009, at 9:55 AM, Hamilton Chua wrote:

 Hello,

 I would like to ask what the plan is for allowing Sugar running on
 hardware other than an XO to register with an XS.

It is encouraged.

 I had a quick look at /usr/bin/registration-server in the XS and
 schoolserver.py in a Sugar installation and I was able to trick  
 the XS
 into registering an SoaS by modifying schoolserver.py on the SoaS. I
 removed the part of the code that checks for the XO information and
 modified the def that handles registration to send randomly generated
 strings for serial and UUID instead.

 I guess it's really not that hard to do but it would be nice to  
 have an
 accepted way of doing it.

 For instance, maybe we could modify schoolserver.py to generate a  
 serial
 and UUID if Sugar is not on an XO laptop instead of saying the  
 required
 info is not present.

Sounds like a great idea!
The only requirement is that serial numbers be unique.  Generating
that in a script can be challenging.   Perhaps a one way hash of the
MAC address ?   But MAC addresses are only unique to one media.

Cheers,
wad


wad
___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Serving 400+ students w/ a single central XS - ejabberd nightmare?

2009-03-10 Thread John Watlington

If you aren't going to place school servers in the actual schools,  
and insist on
centralizing them, the hardware recommended by Sameer is a good idea.

My argument has always been that you want local web caching and content,
and that an XS shouldn't be that much more expensive than the above  
hardware.
If you are willing to guarantee constant connectivity to a central  
office to support
centralized servers, remotely maintaining the servers in the schools  
shouldn't be a problem.

Cheers,
wad

On Mar 10, 2009, at 1:57 PM, Sameer Verma wrote:

 On Tue, Mar 10, 2009 at 3:12 AM, Bryan Berry br...@olenepal.org  
 wrote:
 On Mon, 2009-03-09 at 21:51 -0700, Sameer Verma wrote:
 On Mon, Mar 9, 2009 at 7:04 PM, Bryan Berry br...@olenepal.org  
 wrote:
 On Tue, 2009-03-10 at 09:58 +1300, Martin Langhoff wrote:
 On Mon, Mar 9, 2009 at 12:11 AM, Bryan Berry  
 br...@olenepal.org wrote:
 I am worried about the XO's and not the XS.

 Now you're starting to see what I've seen :-/ I also worry  
 about your
 APs and networking infra -- to support 400 active users you'll  
 want at
 least 8 APs. In more realistic terms, you'll probably need 12,
 assuming a reasonably balanced load.

 We will have roughly 8+ AP's. We have found that off-the-shelf  
 AP's can
 handle around 60-70 users.  But that doesn't still doesn't solve  
 the
 problem of the XO's getting bogged down by tons of ejabberd  
 chatter.

 DSD: do you have any ideas about this?

 We are looking at about 100-150 students per school and  
 connecting 3-4
 schools to a central XS.

 As I mentioned before... I am working on xs-0.6, with the
 moodle-ejabberd magic.

 That's great, but our pilot starts in a month but that doesn't  
 fit our
 timeline. I don't want to send out a completely new, untested XS  
 into
 rural parts of Nepal.

 Do you have any other suggestions fo us?


 What if you had a small footprint box (like a soekris or  
 routerboard)
 at the school that talks to APs on one end via a switch, and does
 tunneling back to XS in a central location? That way you would  
 have a
 fairly dumb tunnel unit at school (literally plug-and-play) and XS
 management back at your central shop.

 Sameer

 Thanks for the suggestion Sameer.

 I don't really understand what benifits the soekris or routerboard  
 adds
 in this situation? Can u pls explain further?


 The Soekris unit (say Soekris 4501 http://www.soekris.com/net4501.htm)
 would sit at the school location talking to the APs via a switch on
 one end and create a tunnel on the other end to your XS farm. The
 tunnel runs over a VPN from school to XS farm. Both Soekris and
 Routerboard have miniPCI slots that will take hardware accelerators
 for VPN such as this one: http://www.soekris.com/vpn1401.htm so it is
 possible to run VPNs on these boards.

 Soekris units are almost zero maintenance (no moving parts etc.) and
 can also double up as APs. One thing to note: the hw VPN accelerator
 support under Linux isn't very robust. I've used it with BSD
 (http://m0n0.ch/wall/) and it works well. It will do IPSec VPN
 tunnels. http://m0n0.ch/wall/features.php

 With a zero maintenance small footprint unit at the school (will run
 with 12 V as well) and XS units at your school server farm you can
 maintain the XS units locally and keep the school network running via
 the VPN.

 The last time I suggested this, the -1 reasoning was that you would
 need a good connection between the school and the XS farm.
 http://lists.laptop.org/pipermail/server-devel/2008-October/ 
 002244.html

 Sameer
 -- 
 Dr. Sameer Verma, Ph.D.
 Associate Professor of Information Systems
 San Francisco State University
 San Francisco CA 94132 USA
 http://verma.sfsu.edu/
 http://opensource.sfsu.edu/ole-nepal- 
 vpn.png___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] rsync on xs-dev

2009-02-12 Thread John Watlington

Currently only through ssh.
I assume you mean an rsync daemon, serving up the xs images ?
I think that could be arranged...

wad

On Feb 12, 2009, at 6:53 PM, Sameer Verma wrote:

 Does xs-dev.laptop.org support rsync?

 Sameer
 -- 
 Dr. Sameer Verma, Ph.D.
 Associate Professor of Information Systems
 San Francisco State University
 San Francisco CA 94132 USA
 http://verma.sfsu.edu/
 http://opensource.sfsu.edu/
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] physical security issue

2008-10-07 Thread John Watlington

You keep pushing for centrally hosted school servers.
Are you sure you don't work for the phone company ?

Again, unless you have a 100 Mbit connection from the
school to the upstream ISP, you will need something with
a disk and a significant amount of memory present in the
school.

I don't disagree about the need for physical security of
the machine, just the proposed solution.

wad

On Oct 7, 2008, at 10:47 PM, Sameer Verma wrote:

 As if discussions on this list aren't lively enough, here's another
 issue to look at.

 While I was in Jamaica, I met with several people who work with their
 school districts, and many pointed out that if a server was to stay
 physically resident at the school, it will need a lot of physical
 security. The most common problem is theft. The other problem will be
 physical damage (just because somebody can). It is not uncommon in
 some of these

 If the school server is hosted at an ISP upstream, we need something
 small (maybe an XO?) at the school that can VLAN or VPN over to the
 school server at the ISP/Data Center.

 Any ideas?

 cheers,
 Sameer
 -- 
 Dr. Sameer Verma, Ph.D.
 Associate Professor of Information Systems
 San Francisco State University
 San Francisco CA 94132 USA
 http://verma.sfsu.edu/
 http://opensource.sfsu.edu/
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] physical security issue

2008-10-07 Thread John Watlington

Actually, Walter, we still hold hope for XOs as school servers
for very small schools.The problem with this is insufficient
memory and insufficient disk space.   While an external disk
may alleviate the second problem, it has poor reliability and
is a very attractive item for theft.

But there is nothing stopping a regular laptop from serving
as a school server.   An external network interface may be
needed for the upstream connection.

wad

On Oct 7, 2008, at 11:25 PM, Walter Bender wrote:

 Clarification: the XO is not the laptop I am proposing for the server.
 Wad can speak to this.

 -walter

 On Tue, Oct 7, 2008 at 11:24 PM, Walter Bender  
 [EMAIL PROTECTED] wrote:
 One idealet (not worthy of being called an idea): What if the server
 were a laptop that the teacher could take with him/her? Pros: The
 school need not be secure. Cons: Price, and of course, laptops can be
 stolen. But it does put the server in the hands of a presumably
 trusted individual in the community.

 -walter

 On Tue, Oct 7, 2008 at 11:20 PM, Sameer Verma [EMAIL PROTECTED]  
 wrote:
 On Tue, Oct 7, 2008 at 8:00 PM, John Watlington [EMAIL PROTECTED]  
 wrote:

 You keep pushing for centrally hosted school servers.
 Are you sure you don't work for the phone company ?


 Last time I checked, San Francisco State University wasn't in the
 telco business.

 Again, unless you have a 100 Mbit connection from the
 school to the upstream ISP, you will need something with
 a disk and a significant amount of memory present in the
 school.


 OK.

 I don't disagree about the need for physical security of
 the machine, just the proposed solution.


 OK. Any other solutions? I'm all ears.

 Sameer

 wad

 On Oct 7, 2008, at 10:47 PM, Sameer Verma wrote:

 As if discussions on this list aren't lively enough, here's  
 another
 issue to look at.

 While I was in Jamaica, I met with several people who work with  
 their
 school districts, and many pointed out that if a server was to  
 stay
 physically resident at the school, it will need a lot of physical
 security. The most common problem is theft. The other problem  
 will be
 physical damage (just because somebody can). It is not uncommon in
 some of these

 If the school server is hosted at an ISP upstream, we need  
 something
 small (maybe an XO?) at the school that can VLAN or VPN over to  
 the
 school server at the ISP/Data Center.

 Any ideas?

 cheers,
 Sameer
 --
 Dr. Sameer Verma, Ph.D.
 Associate Professor of Information Systems
 San Francisco State University
 San Francisco CA 94132 USA
 http://verma.sfsu.edu/
 http://opensource.sfsu.edu/
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel




 --
 Walter Bender
 Sugar Labs
 http://www.sugarlabs.org




 -- 
 Walter Bender
 Sugar Labs
 http://www.sugarlabs.org

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Memory use with SSL connections

2008-10-07 Thread John Watlington

Until Sugar has a better model for collaboration, anything over about
500 users is moot.   (Thanks for testing it all the way up to 2K users).

The problem is that the laptop UI will try to show all the users on the
server in the neighborhood view...   Can you try to connect an XO
to the server with over 500 hyperactivity users, and see what breaks ?

Certainly we are interested in finding and fixing the bottlenecks, but
I don't think pushing past 500 users is critical at this time.

On the other hand, supporting virtual machines so we can have
five ejabber instances running on a server (w. sufficient RAM)
might be an interesting topic to pursue...

Cheers,
wad

On Oct 6, 2008, at 1:07 AM, Douglas Bagnall wrote:

 hi

 I'm testing a slightly modified ejabberd 2.0.1 for OLPC, and am seeing
 quite high memory use -- a 1GB machine is unreliable beyond 2000
 connections.  There are pictures here:

 http://wiki.laptop.org/go/ 
 Ejabberd_resource_tests#Try_4:_a_few_thousand_users

 OLPC's XO laptops use (old style) SSL connections, and these results
 are in the same order of magnitude that Sean Dilda reported in May:

 http://lists.jabber.ru/pipermail/ejabberd/2008-May/003676.html

 For 2,000 SSL connections it'll take around 1GB of RAM.  For 2,000
 plaintext connections, it'll take around 150MB.  I'm seeing similar
 results for the old port 5223 tls and using starttls on port 5222.

 Does anyone have any tips, patches, or configuration options that
 might help?

 regards,


 Douglas Bagnall
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


[Server-devel] resolv.conf

2008-08-26 Thread John Watlington

I've figured out more about why my first install of 0.4 went so badly.

For some reason, the network interfaces weren't configured properly
on first boot (even though the files were in the right places).
In conjuction with having the cable to my wired net plugged into the
LAN port, nothing was happening.

What worried me in addition to the above was that DNS had completely  
broken.
Any DNS queries on the server (host/dig/nslookup), such as  
schoolserver,
were hanging.  But queries directed to the DNS server on localhost  
were just fine.

The answer is that /etc/resolv.conf should probably point to localhost,
not the address of one of the interfaces (172.18.0.1).   Just in case...

I'll git the change after a suitable comment period.

Second install went like a breeze,
wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


[Server-devel] xs-config replacement strategy

2008-08-26 Thread John Watlington

The proposed change to the operation of the xs-config
package seem sane, with a few comments:

The make-a-replacement strategy is the crucial one.
I have hesitations about the xs-config.make file
used:
- Multiple will be needed, in each directory with changed
   files.
- Getting a correct config now requires a correct make file
  and a correct .in file (plus helpers) for each file generated.
- It lumps together changes for multiple packages

It doesn't fix the problem with user changes being deleted ---
the *.in file will change and a new file will be generated
overwriting previous configuration.

What did I miss ?
wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] resolv.conf

2008-08-26 Thread John Watlington

On Aug 26, 2008, at 4:47 PM, Martin Langhoff wrote:

 That sounds like network_config crashed on you. network_config is
 responsible for creating /etc/resolv.conf.in and then domain_config
 will do the rest. On xs-0.4, we call domain_config at install time
 (and it defaults to random.xs.l.o).

I'll try to see if I can figure out why it crashed --- but it created  
the network
config files fine before crashing, if it did...

 The answer is that /etc/resolv.conf should probably point to  
 localhost,
 not the address of one of the interfaces (172.18.0.1).   Just in  
 case...

 Does BIND listen on localhost? If so, it's a good idea...

It most certainly does!

 cheers,
 m

 -- 
  [EMAIL PROTECTED]
  [EMAIL PROTECTED] -- School Server Architect
  - ask interesting questions
  - don't get distracted with shiny stuff - working code first
  - http://wiki.laptop.org/go/User:Martinlanghoff

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] AP with fixed IP address

2008-08-20 Thread John Watlington

I think Jerry misunderstood your question.

You can assign the AP an address from the range 172.18.1.1 to  
172.18.1.254.
This is within the address range assigned to eth1 (in /etc/sysconfig/ 
network-scripts/ifcfg-eth1)
but outside of the range that DHCP is set to provide on that  
interface (.0.2 to .0.254)

Other parameters for machines located statically on the school LAN  
(eth1):
NETMASK  255.255.254.0
GATEWAY 172.18.0.1

Cheers,
John

On Aug 20, 2008, at 6:13 AM, David Leeming wrote:
 Hi Martin,

 To clarify, I am talking about a minimal set up of a school server.  
 Just
 burning the image on a computer that has two network cards  
 installed, and
 following the instructions on the school server wiki page to change  
 the
 domain name and register an ejabberd admin account.

 I am then using a simple access point device like a D-Link 2100AP,  
 set to
 get an IP address from a DHCP server, plugged in to the second  
 network card.
 With this simple configuration, I have found that it then works  
 without any
 more changes to any config files etc.

 My question is, with a particular type of AP that I am required to  
 use,
 which uses a fixed IP, what config changes do I need to make in the  
 XS? The
 AP cannot be set to get an IP address from the XS.

 Is this clear enough?


 David Leeming
 OLPC Coordinator, SPC and Technical Advisor, People First Network
 Honiara, Solomon Islands


 -Original Message-
 From: Martin Langhoff [mailto:[EMAIL PROTECTED]
 Sent: Wednesday, 20 August 2008 10:18 a.m.
 To: David Leeming
 Cc: server-devel
 Subject: Re: [Server-devel] AP with fixed IP address

 2008/8/20 David Leeming [EMAIL PROTECTED]:
 I am setting up a server using a outdoor access point from Rural Link
 (www.rurallink.co.nz) at our Patukae OLPC trial school site in  
 Solomon
 Islands. It's a type that can't get an IP address from a DHCP  
 server and
 on
 the LAN side needs to be fixed.

 Jerry's answer is correct. Edit ifcfg-eth0

 Normally I have found, when setting up the XS, that if I attach a  
 simple
 AP
 to a second NIC using eth1, with DHCP, there is no additional
 configuration
 required. It works by default.

 That sounds wrong. Maybe I'm misunderstanding what you've done but

  - eth0 is the WAN address. Attaching something to eth1 does not solve
 anything unless you've got eth0 and eth1 mixed up...

  - On the LAN side (eth1) don't let the AP act as a router or give
 DHCP leases. Set it to be a vanilla AP, no DHCP, routing or NAT'ting.
 It is up to the XS to do routing, NAT'ting, handing out DHCP leases
 and providing DNS services.

 In the case of my Rural Link access point, I need to fix an IP  
 address
 within the range given by the server - which presumably it also  
 uses to
 allocate addresses to the XOs.

 So you are talking about 2 XSs and want one of the servers to be
 'downstream' of the other? That'll need a bit of routing glue
 methinks. Tell us more about the network topology you're setting up...

 cheers,


 m
 -- 
  [EMAIL PROTECTED]
  [EMAIL PROTECTED] -- School Server Architect
  - ask interesting questions
  - don't get distracted with shiny stuff - working code first
  - http://wiki.laptop.org/go/User:Martinlanghoff

 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Missing XOs in neighbourhood view using XS

2008-07-22 Thread John Watlington

There are two possible reasons.

The first is that you are still using the mesh, and mesh routing with  
that
number of laptops is capricious at best.

The second is a known problem exists which can also cause this:
  http://dev.laptop.org/ticket/6884

We need to get ProcessOne to fix this problem with ejabberd.

Regards,
John

On Jul 21, 2008, at 7:37 PM, David Leeming wrote:

 Trying out the school server for the first time at a training  
 session in a Solomon Islands deployment. With 23 trainee teachers,  
 using G1G1 XO-1 laptops with build 703 (update 1). All registered  
 successfully and all able to connect to and access the server.

 It seemed to be working 100% but then we started noticing that it  
 was not unusual to  be missing some XOs on the neighbourhood view.  
 This was apparent with an exercise where people in groups of 4-5  
 tried to locate each other and make friends, in order to manage  
 sharing in smaller groups. Almost all of the groups could not  
 locate all of the other members of their group.

 I checked, all were registered, connected to the school mesh and  
 all could see most of the other laptops. But not 100% reliably all  
 of the others. Rebooting did not help much.

 This was in a very rural location with no cell phones, no wireless  
 access points apart from the XS active antenna, in a small wooden  
 classroom 6 x 6m. There was, however, an old inverter running; not  
 sure if that might have been making RF noise. We were running the  
 school server from a battery.

 I was using an XS set up to demonstrate the concept only, on a 4- 
 year old Toshiba A3 laptop.  I have a prototype active antenna.

 So (a) should I expect all the other laptops to be visible? And (b)  
 what is the maximum number of XOs per server?

 Server access seems very robust and very impressive. But the  
 unreliable collaboration is a disappointment, it was not much  
 better than a similar size group using the simple mesh. What could  
 be the reason?

 David Leeming
 Honiara, Solomon Islands, South Pacific


___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] eth0 statii IP

2008-07-03 Thread John Watlington

On Jul 3, 2008, at 2:41 PM, Tim Moody wrote:


 - Original Message -
 From: Martin Langhoff [EMAIL PROTECTED]
 To: Tim Moody [EMAIL PROTECTED]
 Cc: server-devel@lists.laptop.org
 Sent: Thursday, July 03, 2008 12:57 PM
 Subject: Re: [Server-devel] eth0 statii IP

The script which performs network configuration (olpc-scripts/network- 
config) is separate
from the script which performs network configuration at first boot  
(init.d/olpc-network-config)
to allow manual operation and simplify debugging.  It is not called  
by any other script.

 On Thu, Jul 3, 2008 at 12:09 PM, Tim Moody [EMAIL PROTECTED]  
 wrote:
 /fsroot.olpc/etc/sysconfig/olpc-scripts/network_config closes the  
 loop by
 creating /etc/sysconfig/olpc_net_config, but ONLY if it  
 succeeds.  It
 will

 Correct. If you change your network hw on the machine, you should run
 the network_config script with a parameter of 1, or remove the
 olpc_net_config flag file and reboot.

 fail if there are more than two wired ethernet adapters (or  
 none); I have
 three so it always fails.

 Ouch. What is the role of the 3rd adapter? Is it valid / useful for
 our core scenarios?

 My machine doubles as a honey wall (2 adapters bridged and 1 to  
 monitor).  I
 don't know if there is a legitimate need for 3 wired adapters in an XS
 scenario, but I have seen a couple of others on this list with the  
 same
 problem (possibly test machines wanting to simulate the mesh over  
 wired

Why would you want to simulate the mesh (outside of doing mesh  
routing alg.
research) ?   Please don't confuse mesh networking with the ability to
collaborate, which works with any network connection between laptops
(WiFi, mesh, wired).  We currently don't recommend that any school  
(except
for ones with less than fifteen students) use mesh networking.   We  
suggest
traditional 802.11g APs, connected to a server over wired ethernet be  
used instead.

 because we can't get an antenna).

Why can't you get an antenna ?   You can request one through our  
developer
program.   Please let me know if you are having problems obtaining  
antennas
through that program --- there have been problems in the past but we  
think
they are fixed!

 Either way, the script could probably accomodate more
 wired adapters.  It already knows what's there.

Sure.  When that script was written, it wasn't clear how long it would
last (I was guessing a month, but it has been almost a year).  If it  
isn't
going to be replaced soon, adding support for three or four adapters
would be easy.

But all adapters beyond the first should be routed subnets, as we are
trying to prevent multicast and broadcast traffic from propagating.
You won't get your honeywall configuration...

 In general a little more flexibility for people who are testing  
 would be
 nice including, while I'm at it, the ability to partition the  
 machine during
 the install.

Disks are cheap.   Why wouldn't we want testers all using the same
partition layout ?   The current partition layout can be critized and
changed, but preserving sanity is easier with a common partitioning
scheme across the XS servers.

BTW, that is a one line change in the kickstart file --- you can  
trivially
build your own school server image that lets you manually set  
partitions...

We need another name for mega-servers, which provide school server
services, but have a different network, service/server mapping, and  
storage
scheme (partitioning or separate disk servers) than the XS.  Suggestions
accepted

Cheers,
wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Target price and quantitiy

2008-06-22 Thread John Watlington

On Jun 22, 2008, at 2:19 PM, Christopher Niemöller wrote:

 Hi,

 I was playing around with some embedded computing and rugged hardware
 and got some questions.

 Whats the target price for the school server?

Around 500 USD for a server plus 500GB disk.
Networking (access points and switches) are additional
(although ideally a four port switch is built in...)

 How many school server are needed anyway?

One per school...   Larger schools need more capable servers.
There is no single solution, from a product design perspective.

If we concentrate on the needs of smaller rural schools (30-300 kids),
we are talking about the $500 server mentioned above, with
tough environmental constraints.  In this case, we are unfortunately
only talking about 5K-10K units per year, and these are spread across
multiple countries.

 As you might know, its my job to design such systems, please dont take
 this questions the wrong way. I dont want to sell anything, but Im
 interested in supporting this project. ;)

We have no problem with people interested in selling stuff.
We need to buy servers ourselves for use in trials, etc...

Cheers,
wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Problems installing XS on new system for Uruguay

2008-06-05 Thread John Watlington

On Jun 5, 2008, at 12:31 PM, Tony Pearson wrote:

 Uruguay = EduBlog/Ceibal/Greg Smith.  I am building the server to  
 Greg's request, so whatever he is now calling this project that is  
 what it is.

Uruguay == Ceibal
Uruguay != EduBlog
Uruguay != Greg Smith

Your statement worries me, as Ceibal has a clear idea of what
they want, and a team of developers working on it.   They also
have a good relationship with their volunteer community.

Why isn't this being coordinated by Ceibal ?

John

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Debian help needed

2008-06-03 Thread John Watlington

Thanks for the offers, but after more discussion with the Uruguay
team, their network architecture is different enough from that on the XS
(all their schools are a single DNS domain!) that the Identity Server
would not work.

They will be developing a simple solution that provides the  
functionality
of the Identity Server as far as the Presence Service is concerned,
and deploying it in their laptop builds.

A longer term solution will require them to rethink how they find the
local school server (right now, they assume that the gateway supplied
by DHCP is the school server.)

Thanks,
wad

On Jun 3, 2008, at 7:42 AM, John Watlington wrote:


 Anybody out there want to help get a school server running on Debian ?
 We need a debian package for the identity manager:

 http://wiki.laptop.org/go/School_Identity_Manager
 http://dev.laptop.org/git?p=projects/idmgr

 (Just to forestall any rumors, this is needed so that Uruguay's school
 servers will support the presence service).

 Cheers,
 wad


___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] uruguay server hardware

2008-06-03 Thread John Watlington

It should go in the server documentation, not networking scenario  
taxonomies.

They do not expect that the winner of the next tender will be IBM with
the same machines.

wad

On Jun 3, 2008, at 6:00 PM, Rafael Enrique Ortiz Guerrero wrote:

 Hi John et all,

 i've wikified this into

 http://wiki.laptop.org/index.php? 
 title=Scenario_taxonomyaction=editsection=5

 maybe there is some better place in the wiki to do so.

 Greg do you agree?. Can we organize this page better?, i'm thinking  
 we should put the specific hardware of all countries in there, to  
 have a knowledge database.



 On Tue, Jun 3, 2008 at 10:47 AM, John Watlington [EMAIL PROTECTED]  
 wrote:

 There has been some questions about what hardware Uruguay is using
 for school servers.
 They are placing a second tender right now, but the first batch of
 servers were:

 (An IBM x3105)
 1.6 - 1.8 GHz AMD processor
 2 GB RAM
 160 GB disk (two drives, in RAID 1)
 three NICs  (one WAN, and two LAN)
 the intent was to have a WiFi network inside the school
and another for outside the school, but they are currently
only using a single WiFi network.
 Six USB slots
 DVD-RW for backup

 These machines stop working at temperatures higher than 40C.
 (Crash, not nice power-off)

 The RAM size was chosen to better support Squid and Dansguardian.

 The small disk size is due to the costs of disks when bought through
 server vendors and there not being XO backup at this point.

 Watchdog and remote monitoring boards are recommended (but not
 currently installed.)

 All errors are mine,
 wad

 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel



 -- 
 Rafael Enrique Ortiz Guerrero
 One Laptop Per Child
 [EMAIL PROTECTED]

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Weird Device Recognition in 163

2008-06-02 Thread John Watlington

On Jun 2, 2008, at 7:16 AM, James Cameron wrote:

 On Mon, Jun 02, 2008 at 08:22:40PM +1000, Martin Langhoff wrote:
 That makes bit more sense. A bad pci connection... but I'm not sure
 why it didn't come up with the right ip addr. I do wonder whether the
 fedora tools have anything to automagically define a device  
 affinity
 once it's seen a given mac address (to avoid logical device swaps if
 you reorder your NICs on a PCI bus). But that's a long shot.

 Hmm.  Eyes sleepy, but
 /etc/udev/rules.d/75-persistent-net-generator.rules if present may
 activate /lib/udev/write_net_rules (script) which may write file
 /etc/udev/rules.d/70-persistent-net.rules ... so if this is what is
 happening there will be evidence.

Perhaps on later Fedora systems.   On the F7 school servers I don't see
75-persistent-net-generator.rules.

Fedora 7 uses a cruder approach, adopted by the XS software, of
placing the device MAC address inside the network script (even if
there is only one device).On first boot, the network-scripts are  
rewritten,
biasing the device assignment toward known reliable NICs as the WAN
port (eth0).

 -- 
 James Cameronmailto:[EMAIL PROTECTED] http:// 
 quozl.netrek.org/
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Weird Device Recognition in 163

2008-06-02 Thread John Watlington

On May 30, 2008, at 5:34 PM, Anna wrote:

 The other weird device thing involved the second NIC. The first  
 one gets recognized as eth0 just fine. However, the second one is  
 seen as dummy0. I put the hardware address for the second NIC in  
 both /etc/sysconfig/olpc-scripts/ifcfg-eth1 and /etc/sysconfig/ 
 network-scripts/ifcfg-eth1, but service network restart still  
 brought up dummy0. As this was late Friday afternoon before a  
 holiday weekend, the library staff was anxious to leave and  
 needed to lock up, so I couldn't play around with it anymore and  
 halted the system. So here's another question - any thoughts on  
 why it wants to see the second NIC as dummy0 and how can I make  
 it be eth1?
 What hw and drivers are loaded?

 I saved off the outputs of lsmod, lspci, dmesg, and /var/log/ 
 messages if you're still interested, but I did get eth1 working.   
 It wasn't being recognized according to lspci, so I popped open the  
 case and moved it to another slot.  Eth1 then initialized on boot,  
 but dummy0 kept getting assigned 172.18.0.1.  So, here's what I did:

 ifdown dummy0
 ifconfig eth1 172.18.0.1 netmask 255.255.240.0
 service dhcpd restart

You've already diagnosed and fixed the problem, but here are some  
more details.

The XS networking setup expects the gateway to be at 172.18.0.1,  
which is normally
assigned to the second wired interface.   If no second wired  
interface can be found
(because the second NIC wasn't being seen), then a dummy0 interface  
is created
with that IP address.

Once you moved the NIC, and it was being discovered correctly, you  
should have been
able to run `/etc/sysconfig/olpc-scripts/network-config 0` (which is  
run at first boot) and it
should have set things up correctly.   That script, however, is not  
well tested for this case
(it expects an unconfigured system).

 The two test XOs were able to pull IPs from the XS after that.   
 Like last Friday, I didn't have time to fool around with it anymore  
 or test registration cause they were closing up the school.

 On my way out, I discovered there's another issue that I'm going to  
 have to find some way to deal with.  The school IT personnel is  
 dead set on finding some way to integrate all the wireless access  
 points and the XS/XO into their existing network.  Why?  I wish I  
 knew.  Anyway, they're on their own IP scheme and they're pitching  
 a fit that the XS isn't in line with that.  172.18 stuff is all  
 over the XS and I can't even fathom changing it.

Until they are willing to take over the server installation and  
maintenance, they shouldn't try to change this.
It isn't hardwired into the XS/XO system per se --- you CAN  
configure a server to work with other IP ranges
inside a school with pre-existing networking services --- but it is  
hardwired into the XS network subsystem.

 I have a real problem with their mentality.  First of all, the  
 existing school infrastructure requires an outside connection, even  
 for basic LAN functionality.  As in, if the connection to the  
 (Windows, naturally) server room downtown goes down, school staff  
 can't even print.  That happens on an appallingly frequent basis.   
 Unless I hear specific, concrete reasons exactly why the XS/XO  
 stuff needs to be integrated into the existing network, I'm going  
 to assume there aren't any and try solidify specific, compelling  
 reasons to not do that.

We will back you up on that.

 My vision, and this is obviously not shared by the school system IT  
 staff (all Windows guys who hate Linux, but don't get me started),  
 is that the XS/XO LANs will be fully independent, self contained  
 ecosystems that doesn't necessarily require outside access to  
 function.  Sure, it'd be nice for the different schools to talk to  
 each other, but it's not vital and if it requires messing around  
 with their existing, unreliable Windows network to get that done,  
 I'd just as soon not have to deal with that headache.  Goodness  
 knows I have an annoying enough time setting up Samba to share  
 files with the XP machine (not mine) on my home LAN.

They hate it because it is unknown and invalidates all their MSCE  
certification.

 Sorry.  Kinda venting here.

No problem.

 Anna Schoolfield
 Birmingham
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Edublog notes (was: Re: The road towards xs-0.3 - update)

2008-06-02 Thread John Watlington

On Jun 2, 2008, at 12:36 PM, Greg Smith (gregmsmi) wrote:

 Hi Martin,

 Thanks a lot for the review. I copied your reply to the Uruguay  
 edublog
 volunteer list: [EMAIL PROTECTED]

 I will use that list to work with end users and for internal project
 tracking. That said, anyone can join. It's Spanish and English and
 occasionally I post in Spanglish :-)

 Tarun is traveling to India this week. Hopefully he or an engineer in
 Montevideo can reply in a few days and we will work out the technical
 details and server related software questions here.

 A few preliminary questions.
 1 - I'm not clear on what you are saying here: DB - assume  
 Postgres 8.x
 series, support mySQL
 Is it PostGres or MySQL? (btw we already brought up a box w/MySQL, so
 hopefully it will be easy to copy tables and queries over).

 2 - Should we have our own table in a single DB that is shared by  
 Moodle
 and all other apps or do need our own DB. I think the DB will be used
 for storing who posted what blog where, pending blogs and all that kid
 of persistent data. Tarun may have other comments on what he needs in
 the DB. We'll do no user auth or identity tracking until we know your
 high level plan. Fall back in case that's not cooked by August is
 cookies or worst case no identity and everyone looks the same.

 3 - We will bring up another XS and will put it on the internet. This
 will be our primary pre-production server (e.g. we may use it for  
 final
 beta test). Let me know if anyone has any concerns about that. Tony  
 who
 worked on Nepal early on will image it and ship it to a co-lo and  
 we'll
 manage it from there. Tony needs the recommended specs:
 HW - memory, capacity disk drives, USB ports, NICs, etc.
 SW - Moodle, Squid, etc. Anything beyond what you get with the XS  
 image
 (e.g. do we need to install Apache and PHP). I'm not sure how we will
 track the F7 dependencies, but I hope the sys admins can think about
 that (still accepting more volunteers too ;-). Tony has notes from the
 first time, so he says its no problem to take Wad's (or Martin's now)
 latest build and go from there.

 The HW and SW specs answer may be RTFM so don't hesitate to send a  
 link.
 I'm asking to get the latest thinking. I'll update the wiki as needed
 once I'm sure I have the right info.

 Also, does anyone have the HW specs of the XS boxes deployed in  
 Uruguay?
 I'll ask elsewhere if no one on this list has it.

I'll try to get that tmw.   I'll be talking to the server team.

wad
___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] XS software compatibility question

2008-05-21 Thread John Watlington

On May 21, 2008, at 1:43 PM, Berkowitz Andrew (Project Connect) wrote:

 OK, the jabber is working fine. The problem was that hostname was
 working correctly but hostname -s  -f were not.

 I installed the idmgr with yum install idmgr.  Are there any
 configuration files I need to edit?

Take a look at /etc/idmgr.conf
It should have the FQDN for the server listed for both variables.

 I did manage to crash the server after updating all packages. I could
 see which one it was.

 This server is running Fedora 7. If I try Fedora 8 or 9 to avoid that
 incompatible package, should your packages still work?

There lie uncharted waters.
Dennis might be able to help with that answer.

I suspect that the answer is probably yes, given the limited
number of packages you are trying.

  Thank you

 -Original Message-
 From: John Watlington [mailto:[EMAIL PROTECTED]
 Sent: Tuesday, May 20, 2008 3:21 PM
 To: Berkowitz Andrew (Project Connect)
 Cc: John Watlington; Martin Langhoff; server-devel
 Subject: Re: [Server-devel] XS software compatibility question


 I don't have enough information to suggest anything, just offer
 condolences and vague comments  :-)

  From your command line info I assume that you are having problems
 keeping ejabberd running, not connecting or collaborating after  
 starting
 it up.

 There is a brief time during startup when ejabberd is running
 but the status is returned as stopped.Other than during time,
 I usually don't see ejabberd processes running while service (or
 ejabberdctl) say it isn't running.  This only lasts for a minute or  
 so.

 Can you take a look at /var/log/ejabberd/sasl.log ?  Usually when
 ejabberd has problems starting, the error is logged cryptically  
 there...

 wad

 On May 20, 2008, at 2:59 PM, Berkowitz Andrew (Project Connect) wrote:

 I am having a problem with the ejabberd.

 I took the ejabberd.cfg  ejabberdctl.cfg files from a working server
 and then changed the domain before starting ejabbered.

 What do you suggest?

 [EMAIL PROTECTED] ~]# ejabberdctl ejabberd status Node
 [EMAIL PROTECTED] is started. Status: started ejabberd is not
 running [EMAIL PROTECTED] ~]# ps auxxwww | grep jabber
 ejabberd  2672  0.0  0.0   1932   340 ?S10:32   0:00
 /usr/lib/erlang/erts-5.6.1/bin/epmd -daemon
 ejabberd  3320  0.1  0.8  65200 18116 ?Sl   10:39   0:00
 /usr/lib/erlang/erts-5.6.1/bin/beam.smp -K true -- -root /usr/lib/
 erlang -progname erl -- -home /var/lib/ejabberd -pa
 /usr/lib/ejabberd-2.0.0/ebin -sname ejabberd -s ejabberd -ejabberd
 config /etc/ejabberd/ejabberd.cfg log_path
 /var/log/ejabberd/ejabberd.log -sasl sasl_error_logger
 {file,/var/log/ejabberd/sasl.log} -mnesia dir
 /var/lib/ejabberd/spool -kernel inetrc /etc/ejabberd/
 ejabberd.inetrc
 -noshell -noinput



 -Original Message-
 From: John Watlington [mailto:[EMAIL PROTECTED]
 Sent: Monday, May 19, 2008 11:39 PM
 To: Berkowitz Andrew (Project Connect)
 Cc: John Watlington; Martin Langhoff; server-devel
 Subject: Re: [Server-devel] XS software compatibility question


 On May 19, 2008, at 6:18 PM, Martin Langhoff wrote:

 On Tue, May 20, 2008 at 9:49 AM, Berkowitz Andrew (Project  
 Connect) 

 The installation was perfect with the full Fedora 7 installation.

 Is it possible to install all of OLPC's customizations on top of a
 stock Fedora installation?

 Yes. Add the XS repositories to your yum sources, and install xs- 
 pkgs

 from there (which should pull all the right dependencies). I suspect
 it will muck up a few things, such as user accts created for
 processes. The kernel might also be missing those specific drivers
 you

 need.

 The config file for ejabberd is in xs-config, so you will probably
 want to replace the Fedora repos with the OLPC ones, and install
 xs-pkgs and xs-config.

 (I'm thinking evil thoughts about just installing the idmgr and
 ejabberd
 packages and grabbing the ejabberd config file from git...   Since in
 the NYC installation you provide all networking services externally!)

 So I can't guarantee it will work.
 Definitely not, but I bet it will be close.
 Don't forget to reboot after installing xs-config, to trigger any
 post-first-boot network configurations...

 How else would you suggest we proceed?

 Hmmm. As you aren't using AAs, you probably don't need ourcustom
 kernel. So perhaps install xs-config and then review what xs-pkgs
 will

 pull in - you'll want to install all of that except for the custom
 kernel.

 We have no custom kernel, our repositories simply contain a freeze of
 the Fedora 7 release/update repos at a recent date.  You are safe
 there...

 Thanks for working on this!
 wad







___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] XS software compatibility question

2008-05-20 Thread John Watlington

I don't have enough information to suggest anything, just offer
condolences and vague comments  :-)

 From your command line info I assume that you are
having problems keeping ejabberd running,
not connecting or collaborating after starting it up.

There is a brief time during startup when ejabberd is running
but the status is returned as stopped.Other than during time,
I usually don't see ejabberd processes running while service
(or ejabberdctl) say it isn't running.  This only lasts for a minute  
or so.

Can you take a look at /var/log/ejabberd/sasl.log ?  Usually
when ejabberd has problems starting, the error is logged
cryptically there...

wad

On May 20, 2008, at 2:59 PM, Berkowitz Andrew (Project Connect) wrote:

 I am having a problem with the ejabberd.

 I took the ejabberd.cfg  ejabberdctl.cfg files from a working server
 and then changed the domain before starting ejabbered.

 What do you suggest?

 [EMAIL PROTECTED] ~]# ejabberdctl ejabberd status
 Node [EMAIL PROTECTED] is started. Status: started
 ejabberd is not running
 [EMAIL PROTECTED] ~]# ps auxxwww | grep jabber
 ejabberd  2672  0.0  0.0   1932   340 ?S10:32   0:00
 /usr/lib/erlang/erts-5.6.1/bin/epmd -daemon
 ejabberd  3320  0.1  0.8  65200 18116 ?Sl   10:39   0:00
 /usr/lib/erlang/erts-5.6.1/bin/beam.smp -K true -- -root /usr/lib/ 
 erlang
 -progname erl -- -home /var/lib/ejabberd -pa
 /usr/lib/ejabberd-2.0.0/ebin -sname ejabberd -s ejabberd -ejabberd
 config /etc/ejabberd/ejabberd.cfg log_path
 /var/log/ejabberd/ejabberd.log -sasl sasl_error_logger
 {file,/var/log/ejabberd/sasl.log} -mnesia dir
 /var/lib/ejabberd/spool -kernel inetrc /etc/ejabberd/ 
 ejabberd.inetrc
 -noshell -noinput



 -Original Message-
 From: John Watlington [mailto:[EMAIL PROTECTED]
 Sent: Monday, May 19, 2008 11:39 PM
 To: Berkowitz Andrew (Project Connect)
 Cc: John Watlington; Martin Langhoff; server-devel
 Subject: Re: [Server-devel] XS software compatibility question


 On May 19, 2008, at 6:18 PM, Martin Langhoff wrote:

 On Tue, May 20, 2008 at 9:49 AM, Berkowitz Andrew (Project Connect) 
 The installation was perfect with the full Fedora 7 installation.

 Is it possible to install all of OLPC's customizations on top of a
 stock Fedora installation?

 Yes. Add the XS repositories to your yum sources, and install xs-pkgs
 from there (which should pull all the right dependencies). I suspect
 it will muck up a few things, such as user accts created for
 processes. The kernel might also be missing those specific drivers  
 you

 need.

 The config file for ejabberd is in xs-config, so you will probably  
 want
 to replace the Fedora repos with the OLPC ones, and install xs-pkgs  
 and
 xs-config.

 (I'm thinking evil thoughts about just installing the idmgr and  
 ejabberd
 packages and grabbing the ejabberd config file from git...   Since in
 the NYC installation you provide all networking services externally!)

 So I can't guarantee it will work.
 Definitely not, but I bet it will be close.
 Don't forget to reboot after installing xs-config, to trigger any
 post-first-boot network configurations...

 How else would you suggest we proceed?

 Hmmm. As you aren't using AAs, you probably don't need ourcustom
 kernel. So perhaps install xs-config and then review what xs-pkgs  
 will

 pull in - you'll want to install all of that except for the custom
 kernel.

 We have no custom kernel, our repositories simply contain a freeze of
 the Fedora 7 release/update repos at a recent date.  You are safe
 there...

 Thanks for working on this!
 wad





___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] XS software compatibility question

2008-05-19 Thread John Watlington

On May 19, 2008, at 6:18 PM, Martin Langhoff wrote:

 On Tue, May 20, 2008 at 9:49 AM, Berkowitz Andrew (Project Connect) 
 The installation was perfect with the full Fedora 7 installation.

 Is it possible to install all of OLPC's customizations on top of a  
 stock
 Fedora installation?

 Yes. Add the XS repositories to your yum sources, and install xs-pkgs
 from there (which should pull all the right dependencies). I suspect
 it will muck up a few things, such as user accts created for
 processes. The kernel might also be missing those specific drivers you
 need.

The config file for ejabberd is in xs-config, so you will probably want
to replace the Fedora repos with the OLPC ones, and install xs-pkgs
and xs-config.

(I'm thinking evil thoughts about just installing the idmgr and ejabberd
packages and grabbing the ejabberd config file from git...   Since in
the NYC installation you provide all networking services externally!)

 So I can't guarantee it will work.
Definitely not, but I bet it will be close.
Don't forget to reboot after installing xs-config, to trigger any  
post-first-boot
network configurations...

 How else would you suggest we proceed?

 Hmmm. As you aren't using AAs, you probably don't need ourcustom
 kernel. So perhaps install xs-config and then review what xs-pkgs will
 pull in - you'll want to install all of that except for the custom
 kernel.

We have no custom kernel, our repositories simply contain a freeze of
the Fedora 7 release/update repos at a recent date.  You are safe  
there...

Thanks for working on this!
wad



___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Fwd: Installing 1.63

2008-05-15 Thread John Watlington

Sorry about that.   The release notes for build 160 mentioned this  
change,
but I stopped mentioning it in later release notes.

If you prefer the noauto build, we can generate it.

wad

On May 15, 2008, at 6:39 AM, Berkowitz Andrew (Project Connect) wrote:

 Yes, I was prompted for the language and time zone via a blue  
 background
 text-mode UI.

 -Original Message-
 From: Martin Langhoff [mailto:[EMAIL PROTECTED]
 Sent: Wednesday, May 14, 2008 8:12 PM
 To: Berkowitz Andrew (Project Connect)
 Cc: John Watlington; server-devel
 Subject: Re: [Server-devel] Fwd: Installing 1.63

 On Thu, May 15, 2008 at 8:19 AM, Berkowitz Andrew (Project Connect)
 [EMAIL PROTECTED] wrote:
   I didn't realize that olpc-install would run automatically.
  The wiki instructions explicitly say that it needs to be run at the
 command prompt.
  Maybe we should fix the wiki.

 I think the issue here is that we've stopped making non-auto images.

 Earlier images were done twice, auto and interactive. Now we are only
 doing autoinstaller images. Did the install process trigger in yours?
 (You get a blue-background text-mode UI, taht starts asking about
 language and tz...)

 cheers,


 m
 --
  [EMAIL PROTECTED]
  [EMAIL PROTECTED] -- School Server Architect
  - ask interesting questions
  - don't get distracted with shiny stuff - working code first
  - http://wiki.laptop.org/go/User:Martinlanghoff


___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Permissions in /sys/class/net/msh0

2008-05-15 Thread John Watlington

On May 15, 2008, at 9:31 PM, Anna wrote:

 Hi All,

 I've been at this for days now and am only emailing out of  
 desperation.

Hi, Anna.   Good to hear from you again!
Please don't wait until you are desperate to drop us a note.

 I'm on server build 161 (the 163 install didn't work on my hardware),
 though I enabled the testing repos to get what I hope is closer to  
 163.

What was the error when installing 163 ?   We want to know about  
these things!

 Anyway, I finally managed to trick the server into seeing a msh0
 device.  I've got 3 NICs - eth0, eth1, and msh0.  I put HWADDR=mac
 address for the third NIC in the ifcfg-msh0 files.

Why did you have to trick it ?   Do you have an Active Antenna ?

 Now that I've got /sys/class/net/msh0, the olpc-mesh-script wants  
 to run
 at boot.  Which is great!  But, something's the matter with the
 permissions.  On boot, when it tries to echo 1 
 /sys/class/net/msh0/anycast_mask, permission is denied.  I can't  
 even do
 it manually.

When I've seen this in the past, it was because the driver wasn't the  
right
one...   Unless the driver for the third interface is the libertas  
driver, this will fail.

 I'm guessing there's a udev rule somewhere that needs to be edited or
 added.  Any ideas?

 The reason I'm doing this is to try to see if I can get a school mesh
 portal point to work with off the shelf hardware.  If anyone has an
 easier way to do that or can point me towards the documentation, I  
 would
 be grateful for any assistance.

I don't think there is any way currently to run a (pseudo) 802.11s  
mesh with
OTS 802.11b/g hardware at this time, although the Open802.11s project is
working on that.   If you have OTS APs, just hook them up to the  
second NIC
and let the laptops connect that way.

We do have Active Antennas available.   Send mail to  
[EMAIL PROTECTED]
and specifically request an Active Antenna.   Please cc me, as the  
process
by which these are supposed to get fulfilled has been broken lately...

Cheers,
wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Permissions in /sys/class/net/msh0

2008-05-15 Thread John Watlington

On May 15, 2008, at 10:56 PM, Martin Langhoff wrote:

 On Fri, May 16, 2008 at 2:50 PM, John Watlington [EMAIL PROTECTED]  
 wrote:
 Ouch.  That sounds familiar from early days of debugging the install.
 It sounds like your Frankenmachine is a 586, and needs the restricted
 kernel I put in build 161.  The next major release should probably
 include a 586 version as well...

 Should we ship all the regular Fedora kernels there, and compile the
 libertas module against each of them?

Sounds great.  Bet we can't do it easily using livecd/anaconda...

 From our last discussion on kernel builds, my impression was that I
 would be better off building libertas against the stock Fedora kernels
 and packaging it into its own rpm.

Dennis was supposed to be working on this.  Dennis ?

 The bad news is that backup is still being implemented...  It is on
 the high priority list for the next release of laptop software
 (8.2, due out in August).   We have scripts that you can try with a
 school server now, for disaster recovery purposes.

 Indeed - working on it right now.

Great.   What approach are you taking (new thread, perhaps ?)

wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Possible Jabber Problem on Server

2008-05-14 Thread John Watlington

Nothing but ejabberd needs to be uninstalled.   Use the --nodeps
to force this.

Just reinstalling the package doesn't rebuild the database.
You will have to manually clean out the database before
reinstalling...

I have no experience with uninstalling xs-pkgs and xs-config.
I doubt it would reinstall cleanly...

wad

On May 14, 2008, at 8:45 AM, Berkowitz Andrew (Project Connect) wrote:

 I had already uninstalled ejabberd  xs-pkgs. xs-config also had to be
 uninstalled and reinstalled. So I had to redo a few config files prior
 to starting ejabberd. That has always worked just fine. But it took
 about ten minutes altogether.

 I uninstalled and reinstalled a few times to make sure I could do  
 so as
 quickly as possible.

 I didn't realize it would be safe to specify nodeps. If I had, there
 would have been a lot less work to reconfigure the server.

 I hoped that following your suggestion would be a bit less work. We  
 have
 about 1500 schools, so I am searching for the most efficient server
 management processes possible.

 So after reinstalling jabber and getting it behaving again, I tried
 following your suggestion and just deleting the files. That did not
 work. So I reinstalled again. This process is what I was asking a  
 bout.

 If we wind up with a lot of servers, I'll put together a Perl-based
 server management system.

We certainly expect to have a lot of server, and remote management
is one of the tasks that has to be handled.This wasn't remote
management so much as a bug fix.   We have several issues with
ejabberd, and in the long term are seeking a replacement instead
of finding work-arounds.

 -Original Message-
 From: Martin Langhoff [mailto:[EMAIL PROTECTED]
 Sent: Wednesday, May 14, 2008 12:30 AM
 To: Berkowitz Andrew (Project Connect)
 Cc: Server-devel@lists.laptop.org
 Subject: Re: [Server-devel] Possible Jabber Problem on Server

 On Wed, May 14, 2008 at 5:36 AM, Berkowitz Andrew (Project Connect)
 [EMAIL PROTECTED] wrote:
  Martin had suggested that we can just wipe out the database without
 reinstalling the package.

  So back to the question: Can the database be wiped out and recreated
 without reinstalling?

 We haven't tested it :-/  (it all depends on exactly what triggers an
 empty DB setup).

 OTOH, we know that uninstalling/reinstalling the rpm does the trick
 reliably (so it is better than my earlier suggestion). It will remove
 the xs-pkgs package, but that won't cause any short-term problem.
 Reinstall it as well just in case for better long term behaviour ;-)

 cheers,


 m
 --
  [EMAIL PROTECTED]
  [EMAIL PROTECTED] -- School Server Architect
  - ask interesting questions
  - don't get distracted with shiny stuff - working code first
  - http://wiki.laptop.org/go/User:Martinlanghoff
 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


[Server-devel] Fwd: Installing 1.63

2008-05-14 Thread John Watlington

Can anybody suggest the cause for this ?
I haven't seen this problem before.

wad

Begin forwarded message:

 From: Berkowitz Andrew (Project Connect)  
 [EMAIL PROTECTED]
 Date: May 14, 2008 11:14:11 AM EDT
 To: John Watlington [EMAIL PROTECTED]
 Cc: Martin Langhoff [EMAIL PROTECTED]
 Subject: Installing 1.63

 I just tried loading 1.63 onto an old Dell PowerEdge 2400 for another
 test server.

 I receive the following:
 Loading vmlinuz
 Loading initrd.img
 ...
 Ready.
 Vmlinuz: attempted DOS system call
 boot:

 It waits for me to type something at the prompt.

 What should I type at the boot prompt?

 Thank you.



___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Possible Jabber Problem on Server

2008-05-13 Thread John Watlington

I was very explicit in my instructions:
1) Remove the package.
2) Remove the database.
3) Install the package.

Installing the package probably rebuilds the database.
If you re-install without removing the old database, the
re-install seems to re-use the old database.

John

On May 12, 2008, at 1:55 PM, Andrew Berkowitz wrote:

 Should ejabberd automatically create a new database if it does not  
 find one?

 I had already tried deleting the files. But then ejabberd wouldn't  
 start.

 Andrew Berkowitz
 [EMAIL PROTECTED]
 NYCBoE: 718-935-5471
 cell: 917-613-3941

 On May 8, 2008, at 5:15 PM, Andrew Berkowitz wrote:

  Or, how do I just wipe out the mnesia database and start over?
 
 I believe
 rm -r /var/lib/ejabberd/*


___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


[Server-devel] Build 163

2008-05-08 Thread John Watlington

Build 163 has finally been smoke tested on a few platforms
and released.  This should be used for any new installations.
This is a bug fix release of build 160, to ensure that ejabberd
collaboration works properly when the school server can't be
resolved using the DNS root servers (most cases !)

http://wiki.laptop.org/go/XS_Installing_Software#OLPC_XS_161
or download directly from:
http://xs-dev.laptop.org/xs/OLPC_XS_163.iso

Build 161 had libertas driver problems, and an improper
fix to the above problem.

I will push these packages (currently in the testing repo) into the
stable repo sometime in the next week, probably incorporating
a few more of Martin's bug fixes..

We of course eagerly await Martin's first official release!

A note about Active Antennas:  I was unable to test collaboration
using a mesh at 1CC.   Laptops could DHCP, and register, but
failed to connect with the ejabberd server (eventually dropping
to using salut, or link-local XMPP).   This was probably due to
the large number (over twenty) of other laptops in the area
on the same channel, which weren't registered to this new, test
server.  The same server worked fine when moved to an
environment with only three laptops and tested.
The moral is to always keep your laptops registered with the
server, to avoid problems...  Time to automate this ?

Cheers,
wad
___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Possible Jabber Problem on Server

2008-05-08 Thread John Watlington

On May 8, 2008, at 5:15 PM, Andrew Berkowitz wrote:

 Or, how do I just wipe out the mnesia database and start over?

I believe
rm -r /var/lib/ejabberd/*


___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Possible Jabber Problem on Server

2008-05-07 Thread John Watlington

On May 7, 2008, at 10:28 PM, Martin Langhoff wrote:

 On Thu, May 8, 2008 at 2:19 AM, Berkowitz Andrew (Project Connect)
 Greetings from New York.

 G'day mate from New Zealand!

  Although the short hostname, schoolserver, has stayed the same,  
 the domain
 has changed from 00b000.nycboe.org to 00b001.nycboe.org and back to
 00b000.nycboe.org. When I changed the domain  address, I changed the

 H. That's unsupported at the moment. Wad might have more
 experience changing the FQDN...

I've only recovered from that by nuking the ejabberd installation
(rpm -e ejabberd; rm -r /var/lib/ejabberd/; yum install ejabberd)

 Jabber servers are
 notoriously finicky and frequently return the nodedown message if  
 a server's
 hostname changes.

 Exactly, and we haven't yet done the legwork to make sure that this
 works end-to-end.

We've done enough to know that changing the host/domain name is a big  
problem
with the current presence server.

 The normal resolution is to export the database, convert it and  
 import it. I
 tried this conversion using the Erlang conversion utility.

 I also reimported the original database. Some people report that just
 exporting and importing, without conversion, fixes the problem.  
 This process
 also did not resolve the issue.

I found at least one recipe which claimed to fix this problem a while  
back, but didn't test it.

 Hm. What happens if you wipe out the mnesia database and start from  
 scratch.

Fixes the problem.

  Also, I have another concern. The jabber database is kept in text  
 files.
 This configuration is fine for small sites. But with many users,  
 it is
 better to keep the database in MySQL for better performance and  
 better
 reliability.

 If I can get mnesia to use PostgreSQL as a backend, I'll do it.

There are configuration file option for ejabberd to use postgreSQL, etc.
for the database.   Perhaps some volunteers can play with these ?

 Another question: You've reported a bug where If an XO is reimaged  
 and then
 needs to re-register with the server, it cannot. In this case, is  
 it okay to
 export the database as text, manually delete the entry, and then  
 import the
 database.

 For a quick fix, you can delete the entries in the SQLite db in
 /home/idmgr . We'll have a UI for this later.

Wrong database.  If that bug is due to a database, it is the ejabberd
database.

But if you are referring to #6919, it is harder to replicate than  
Giannis indicated
at first.   I do this all the time, when testing builds/servers, and  
hadn't seen it.
It was easier to replicate for a while, when the presence server in  
the local area
was misconfigured.

 All these ad-hoc DBs need to be under some control, and the UI will
 probably be handled as an extension of Moodle's user mgmt facilities.

Darn right...

Cheers,
wad


___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Odd Libertas load error

2008-05-06 Thread John Watlington

On May 6, 2008, at 4:12 PM, Martin Langhoff wrote:

 On Wed, May 7, 2008 at 1:38 AM, John Watlington [EMAIL PROTECTED]  
 wrote:
  Dennis correctly debugged the problem.   This is due to the
  attempt to produce a build for 586 machines, hoping to
  run on the VIA C3.

 that's great to hear - though I have to confess that I'd like to see
 the build process (and to be able to repro it). I had forgotten that
 161 had a different kernel from 160 with the ViaC3 changes.

Build process for 162:

I went into the xs/testing/updates/7/i386 repo, restored the 686  
versions of the
kernel, and did a createrepo

I made the changes I wanted to xs-config, pushed them back onto git.l.o,
built the package on xs-dev, copied it into xs/testing/olpc/7/i386  
repo and did a createrepo.
http://dev.laptop.org/git?p=projects/xs-config;a=summary

On duplicator.l.o, I built the livecd.  I rebooted duplicator on  
noticing that /dev was munged,
then copied the ISO back onto xs-dev for distribution and burned it  
onto a USB key for local testing.

  I'm building 162 right now, which should fix this problem.
  I became aware of it last night.   I guess all the build 161
  installs I did were AP based...

 H. Perhaps at 1CC I was testing mainly with 160? I'm sure I got
 161 working at some point, or perhaps I had 160 + new packages. In any
 case, I think I pushed out some xs-config changes.

Umm, yes.   git.l.o doesn't show any check-ins by you (and I pulled  
before starting
my changes).  I built release 2, copied it into the repo, not knowing  
that you had
already bumped up to release 3...   I'm rebuilding 163 w. release 4,  
and will rebuild
with release 5 when you merge and commit your changes...

I've installed and used build 161 on at least four servers (AMD and  
Intel), and they
all worked fine through various loopholes in the problems  
(domain_config worked
OK if the machine hadn't DHCP'ed, if using APs the libertas driver  
wasn't loaded, etc..)

QA !I already know to test on at least two wired network  
configurations, now we need
to start ensuring tests in both AP and mesh mode.

  (Already checked into xs-config 2.7.1r2)   I had a script which
  over-wrote /etc/resolv.conf after dhclient wrote it, but it was
  erroneously keeping some of the DNS servers.  Instead of
  fixing it, in build 161 I deleted it...

 I'll look into that one. Once dhclient writes its resolv.conf, we need
 to write a forwarders { $dnsservers }; forwarders only;  config for
 BIND.

I would suggest not using forwarders only unless necessary (the
VSAT case for this in Peru was static IP assignment.)   If the  
upstream DNS
goes down, the school server is perfectly capable of working down  
from the roots.

Debian has the resolvconf package, which does this, or we can cook  
our own.
We also want to update the provided external address in the  
nameserver, and
provide the update to an upstream nameserver...

  I'll do some smoke testing of 162 this morning.

 Cool. I'm on IRC.

 cheers,

 m
 -- 
  [EMAIL PROTECTED]
  [EMAIL PROTECTED] -- School Server Architect
  - ask interesting questions
  - don't get distracted with shiny stuff - working code first
  - http://wiki.laptop.org/go/User:Martinlanghoff

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Odd Libertas load error

2008-05-01 Thread John Watlington

On May 2, 2008, at 12:57 AM, Martin Langhoff wrote:

 Testing installation of a new build today with a hand-me-down Shuttle
 machine I got a strange error from the Libertas driver.

 libertas: version magic '2.6.23.1-21.fc7 SMP mod_unload 686 4KSTACKS '
 should be '2.6.23.1-21.fc7 SMP mod_unload 586 4KSTACKS '
 usb8xxx: version magic '2.6.23.1-21.fc7 SMP mod_unload 686 4KSTACKS '
 should be '2.6.23.1-21.fc7 SMP mod_unload 586 4KSTACKS '

 I repro'd this with 161, and (just in case) I tested downgrading the
 firmware to the 160 version (usb8388-5.110.20.p49.bin). This is a P4
 1.80GHz, I wonder if we are compiling the libertas driver with 686
 flags?

I'll get back to you on that.   This isn't the firmware, it is the
/lib/modules/`uname -r`/kernel/drivers/net/wireless/libertas/*.ko
modules, libertas and usb8xxx
Right now the horrible inclusion path is manually pulling it from the
olpc-2.6/stable tree, patching it for the 2.6.23 kernel, and checking  
the
compiled modules into the xs-config git tree.   (Traditionally done on
schoolserver.laptop.org, not xs-dev.)   Breaking this out and automating
it is long overdue.

Cheers,
wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Extreme Linux Server Available to North America

2008-04-30 Thread John Watlington

On Apr 29, 2008, at 7:00 PM, Martin Langhoff wrote:

 2008/4/30 Holger Levsen [EMAIL PROTECTED]:
  On Tuesday 29 April 2008 22:41, Martin Langhoff wrote:
 If those same machines could be ordered with RAM ranging from 512 to
 2GB... instant love ;-) -- add a fanless ext USB drive enclosure and
 run run run.

  did you contact the manufacturer?

 No. But if there are higher spec'ed machines in the pipeline, it'd be
 good to hear about them.

Yes, Peru is in contact and I will try to start a direct discussion.

John

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] Network Provisioning

2008-04-25 Thread John Watlington

On Apr 25, 2008, at 8:29 AM, Stefan Reitz wrote:

 Hi Y'all,

 On Fri, 2008-04-25 at 14:57 +1200, Martin Langhoff wrote:

 On Fri, Apr 25, 2008 at 2:38 PM, John Watlington  
 [EMAIL PROTECTED] wrote:
 
  Proposed change to the hardware spec:
 
  From one to four access points may use an simpler switch,
  connected to the server over a 100 Mb/s link. From five to seven
  access points will need a better switch, which provides a 1 GB/s
  link to the server.
 
   This means that a 1 GB/s interface should be specified for the  
 servers.

 I feel teleported back to my 3-19-08 mail:

 [...]
 (I am thinking  server clusters - red hat cluster suite is offering  
 some nice tools which I never got a chance to use / try - thinking  
 3 servers with fail-over and increased performance for clients  
 (like two servers actually doing something...) would be a starting  
 point)
 Birmingham is looking at 49 schools with a total 14,000+ students.
 [...]


 Theoretically, yes... but perhaps this is a bit over the top. For the
 space we are aiming...

 please define our aim

Martin was correct in that the aim of this discussion is rural  
schools in
Peru.   Birmingham, NYC, and others will require heftier servers.

I do dispute any claim that 1Gb/s network interfaces are over the top at
this point in time.   The cost difference on the manufacturing side is
around $2.

 - the XS services will bottleneck well before saturating 1Gb/s  
 traffic
 - 'upstream' services that the XS is routing will bottleneck well  
 before 1Gb/s

 if we see a 7-AP setup, it will be there to support either a large
 number of laptops or a location with obstacles that needs many
 antennaes. In any case, it will support laptops mostly peering w
 eachother.

 how about those 14 - 28 AP setups?


 If we are designing for a client base of laptops that we actually
 expect to saturate 1Gb, then... we need to start recommending a
 mid-range server cluster, perhaps a SAN, all costing a few megabucks
 --   http://en.wikipedia.org/wiki/Weasel_words

 ;-)


 You get pretty decent off-the-shelf hardware for $500 (asus m2n,  
 athlon x2 xx , 2GB RAM) + cost of storage space) per server.
 And considering not every deployment is a remote underelectrified  
 mountain / desert / ... area, this should not be light heartedly  
 dismissed.

 cheers,

 m

 And I agree with Aaron Huslage that the nature of the AP is going  
 to be another big hitter. But I really haven't seen the numbers on  
 the budget yet. A decent (non-WRT54(...)) AP comes for $300 - 450  
 and is still worth considering.


Please define decent, and how it differs from cheap.   We have run  
into APs that appeared to artificially cap the number of connections  
to less than 30 (market segmentation ?), but have also tested $50 APs  
which seem to support 50 users fine.

Yes, centrally managed networks of APs are much better.
Is that the $300-$450 price you quoted, and does that include the  
controller ?

wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


[Server-devel] Fwd: Network Provisioning

2008-04-24 Thread John Watlington

On Apr 24, 2008, at 10:57 PM, Martin Langhoff wrote:

 On Fri, Apr 25, 2008 at 2:38 PM, John Watlington  
 [EMAIL PROTECTED] wrote:

  Proposed change to the hardware spec:

  From one to four access points may use an simpler switch,
  connected to the server over a 100 Mb/s link.   From five to seven
  access points will need a better switch, which provides a 1 GB/s
  link to the server.

  This means that a 1 GB/s interface should be specified for the  
 servers.

 Theoretically, yes... but perhaps this is a bit over the top. For the
 space we are aiming...

  - the XS services will bottleneck well before saturating 1Gb/s  
 traffic
  - 'upstream' services that the XS is routing will bottleneck well  
 before 1Gb/s

 if we see a 7-AP setup, it will be there to support either a large
 number of laptops or a location with obstacles that needs many
 antennaes. In any case, it will support laptops mostly peering w
 each other.

Wrong.   Right now all collaboration moves through the ejabberd server.
We hope to change that, but it won't happen for roughly a year.

 If we are designing for a client base of laptops that we actually
 expect to saturate 1Gb, then... we need to start recommending a
 mid-range server cluster, perhaps a SAN, all costing a few megabucks
 ;-)

But a school of 250 students will need at least five access points.
It only takes two laptops to saturate a channel (OK, maybe one).
So you are saying that squid or apache can't keep up with feeding
ten streams at 11+ Mb/s each ?

wad


___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


[Server-devel] Collaboration between schools

2008-04-23 Thread John Watlington

I learned more about the network built by the MED in Peru for their
schools.   Each school is in its own VLAN, and cannot route to the
other schools, only to the Internet and to MED servers.

They have good economic reasons for encouraging this, but
it means that inter-school collaboration will have to happen
through data pushed to an MED server (and won't be real-time
activity collaboration).

wad

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] AP association, and mesh behaviour

2008-04-20 Thread John Watlington

And has been directly reported as #6872

On Apr 18, 2008, at 5:11 PM, John Watlington wrote:


 This problem is believed to be the same as #5459.
 http://dev.laptop.org/ticket/5459

 On Apr 16, 2008, at 10:23 AM, sulochan acharya wrote:

 Do you know whether it's just a UI thing, or does it mean that the
 antenna is still scanning? How can we tell if it is? (and is the
 scannign costly resource-wise?)

   You are wrong about its behavior with known APs.   The laptop  
 always
   searches for mesh portals.  If it doesn't find any, it then  
 looks for
   APs that it knows about.  If it finds one, it selects it.



 -- We've seen similar behaviour here in Nepal. I was believing  
 that this is only an UI problem and not network as a whole. I  
 didn't find anything unusual in wireshark either or maybe missed  
 it completely. I've seen this with Q2D14 and 656, 702 and 703.

 Is there a way to completely disable the mesh search and get the  
 XO to display only the networks available?? We are going for  
 regular wifi as AP, and mesh portal search seem to add confusion.

 best,
 Sulo


___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] AP association, and mesh behaviour

2008-04-18 Thread John Watlington

This problem is believed to be the same as #5459.
http://dev.laptop.org/ticket/5459

On Apr 16, 2008, at 10:23 AM, sulochan acharya wrote:

 Do you know whether it's just a UI thing, or does it mean that the
 antenna is still scanning? How can we tell if it is? (and is the
 scannign costly resource-wise?)

   You are wrong about its behavior with known APs.   The laptop  
 always
   searches for mesh portals.  If it doesn't find any, it then  
 looks for
   APs that it knows about.  If it finds one, it selects it.



 -- We've seen similar behaviour here in Nepal. I was believing  
 that this is only an UI problem and not network as a whole. I  
 didn't find anything unusual in wireshark either or maybe missed it  
 completely. I've seen this with Q2D14 and 656, 702 and 703.

 Is there a way to completely disable the mesh search and get the XO  
 to display only the networks available?? We are going for regular  
 wifi as AP, and mesh portal search seem to add confusion.

 best,
 Sulo

___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel


Re: [Server-devel] GSoC: XS admin interface design plan

2008-03-23 Thread John Watlington

This combines monitoring with configuration.  Perhaps a clearer split  
at the top ?
Some monitoring should be open (as long as privacy considerations are
taken into account).

I see no need to configure iptables.   We are open by default, we  
only run
iptables to redirect for transparent proxy and protect one open port on
ejabberd right now.

What parameters would you allow tweaking for HTTP caching ?  I see zero.

You forgot WAN, tunnel, NAT, and DNS configuration.
All interfaces should request IPv6 as well as IPv4 information as  
appropriate.

Cheers,
wad

On Mar 24, 2008, at 2:32 AM, crosvera wrote:

 Hi people:

 In my last mail, I wrote that I would like design the XS's admin
 interface, so I'm making a plan to build it, here is:
 http://crosvera.googlepages.com/plan

 I would like get your feedback to improve the plan.


 Cheers :)
 -- 
 Carlos Ríos V.
 Estudiante de Ing. de Ejec. en Comp. e Inf.
 Universidad del Bío-Bío

 ___
 Server-devel mailing list
 Server-devel@lists.laptop.org
 http://lists.laptop.org/listinfo/server-devel


___
Server-devel mailing list
Server-devel@lists.laptop.org
http://lists.laptop.org/listinfo/server-devel