Zotonic
Zotonic
zotonic@conference.zotonic.com
Tuesday, 16 April 2013< ^ >
arjan has set the subject to: Zotonic - the Erlang Content Management Framework
Room Configuration

GMT+2
[04:45:11] Jarimatti joins the room
[06:09:53] arcusfelis joins the room
[06:24:31] Jarimatti leaves the room
[08:00:20] arcusfelis leaves the room
[09:04:56] Arjan joins the room
[09:24:43] arcusfelis joins the room
[09:33:17] Maas joins the room
[09:43:11] Andreas Stenius joins the room
[09:55:26] Andreas Stenius leaves the room
[09:58:16] <Marc Worrell> That clock was too strange. Funky.
[10:01:22] Andreas Stenius joins the room
[10:01:38] <Andreas Stenius> yep
[10:01:42] <Maas> I wonder, was ntpd running on that server?
[10:01:55] <Marc Worrell> but 5 minutes????
[10:01:57] <Andreas Stenius> I don't recall seeing that process..
[10:02:09] <Marc Worrell> maybe two ntpd servers fighting over the clock?
[10:02:10] <Andreas Stenius> I had the same idea, but it was getting late..
[10:02:24] <Maas> :-) Probably a bad crystal on the mobo.
[10:02:45] <Andreas Stenius> yeah, but with different sources? that disagreed on the time by exactly 5 minutes!?
[10:02:49] <Marc Worrell> that would explain clock drifting, but not jumping, wouldn't it?
[10:03:20] <Andreas Stenius> simon has directed his questions regarding this to his VPS hosting company..
[10:03:38] <Andreas Stenius> will be interesting to hear what they have to say about it
[10:03:55] <Maas> Jumping is odd indeed, and ntpd does not set back the clock, only slows it down.
[10:04:28] <Andreas Stenius> yeah, so every 2-3 minutes it would skip ahead five minutes, for a few minutes, then drop back again..
[10:05:32] <Andreas Stenius> it did show those symptoms early on in a test we asked simon to run, but he overlooked it (on april 5:th)
[10:06:42] <Andreas Stenius> I'm quite relieved it had nothing to do with zotonic nor erlang.. :)
[10:08:21] <Marc Worrell> yeah - jumping clocks will make any time based system crazy
[10:08:34] <Marc Worrell> maybe something to do with the VPS software
[10:08:42] <Maas> I was a bit worried about the linux time thing I saw in the erlang code. They just used normal system calls to get the time, nothing strange there.
[10:09:11] <Maas> Only for the cludge they did some ad hoc bit shifting.
[10:11:37] <Maas> faulty hardware can be very strange.
[10:12:21] <Marc Worrell> http://serverfault.com/questions/451150/time-jumps-on-guest-system-really-shortly
[10:15:31] <Marc Worrell> What might be is that NTPD sets the time correctly, but that Xen (or VMWare or whatever) sets it back/forward 5 minutes, as the hardware is 5 minutes off?
[10:16:18] <Maas> Sigh.. a whole new level of configuration errors...
[10:17:28] <Marc Worrell> http://communities.vmware.com/message/861868
[10:17:35] <Marc Worrell> Add 'clock=pit' to your kernel boot parameters in your guest's boot loader.
 
 
For example in Grub'g menu.lst:  kernel /boot/vmlinuz root=/dev/sda1 vga=791 resume=/dev/sda2 clock=pit
 
 
That usually clears the problem for Linux guests.
[10:17:48] <Marc Worrell> This is really too crazy
[10:18:15] <Marc Worrell> One more reason to like real hardware - at lest you know what you are up to...
[10:19:33] <Maas> Yep, this is really bad.
[10:21:45] <Maas> They say this in the documentation: Do not configure the virtual machine to synchronize to its own (virtual) hardware clock, not even as a fallback
with a high stratum number.
[10:22:54] <Maas> Funny: If the guest operating system is using tick counting and it does not get enough CPU time to handle the number of
timer interrupts per second that it has requested, its clock falls behind real time.
[10:27:11] <Maas> Better check our vm setup. :-) Xen
[10:27:47] <Marc Worrell> indeed
[10:28:10] <Marc Worrell> so always use the software clock and some ntpd sauce
[10:28:42] <Andreas Stenius> I like tomato :p
[10:28:52] <Maas> If you're running a Linux kernel where /proc/sys/xen exists, you need to disable synchronization with the host, so that it only uses ntpd as a time source:
[10:28:55] <Marc Worrell> a real pizza box :p
[10:29:04] <Andreas Stenius> yay, and cheese ;)
[10:29:33] <Marc Worrell> Quatorze timeori pizza-tomato :p
[10:31:19] <Maas> cinque minuto saltare
[10:32:15] <Andreas Stenius> no parla italiano, si?
[10:39:30] <Maas> our vm was configured correctly by default
[10:44:03] <Marc Worrell> continuing with rabbitmq struggle (another project) :)
[10:44:40] <Maas> Almost sounds like the config woes a client of me had once with ibm mq series... They where loosing messages, while they just bought the mq series to not let that happen...
[10:46:01] <Maas> One of the queue managers was configured to drop messages after some time. That was a big shock for them that you could configure it to loose messages.
[10:47:25] <Maas> I guess rabbitmq means a lot of configuration hassle too.
[10:49:15] <Maas> When are we going to spec our pub/sub thingy? :-p
[10:52:11] <Marc Worrell> really soon now :)
[10:52:22] <Marc Worrell> I looked a bit more in the mobwrite thing
[10:52:35] <Marc Worrell> looks good - really good for mobile as well
[10:53:10] <Andreas Stenius> not sure if this is what usually trips people up with the name issue, but I had to add my hostname to /etc/hosts and then it worked, so it was dead simple...
[10:53:12] <Marc Worrell> And the C++ diff/match/patch is easy to get working - so we can prob make it a NIF (and some Erlang code for the easy insert/delete cases)
[10:57:59] arcusfelis leaves the room
[10:59:04] <Marc Worrell> O'Reilly 50% of ebooks for the coming week, discount code: WKESEVE
[10:59:21] <Marc Worrell> http://shop.oreilly.com/category/deals/elsevier.do?code=WKESEVE
[10:59:25] simon.smithies joins the room
[11:07:36] <Maas> Nice, when did they start selling non-oreilly e-books?
[11:18:38] <Marc Worrell> no idea, but they added Elsevier just now
[11:21:24] <Maas> It would be really cool if they could add mit press too.
[11:36:07] simon.smithies leaves the room
[12:36:45] arcusfelis joins the room
[14:14:01] <Maas> weird, doesn't android have websocket support in its browser...
[14:14:29] <Maas> if("WebSocket" in window) {
console.log("websocket connect");
do_ws_connect();
} else {
console.log("comet connect");
do_comet_connect();
}
[14:14:49] <Maas> Android goes for comet..
[14:15:30] <Arjan> http://caniuse.com/websockets
[14:17:02] <Marc Worrell> Android is primitief :p
[14:17:20] <Maas> comet also fails :p
[14:17:24] <Marc Worrell> oh?
[14:17:43] <Marc Worrell> why? that is a normal long poll
[14:17:49] <Maas> domain and protocols and ports must match... and they are the same
[14:17:59] <Maas> but still it fails.
[14:18:14] <Marc Worrell> aha, they are the same so they are not the same so it should work but fails
[14:18:21] <Marc Worrell> sounds logical
[14:18:29] <Marc Worrell> *needs food*
[14:18:41] <Maas> domain name is this: c.fritz.box:8000
[14:19:29] <Marc Worrell> completely ok
[14:20:29] <Marc Worrell> food - and fetching Mr T :)
[14:20:35] <Maas> sigh.. one of the urls is missing the protocol
[14:21:07] <Marc Worrell> oh? our urls?
[14:21:16] <Marc Worrell> or a config thing?
[14:21:19] <Maas> My bus :-)
[14:21:26] <Marc Worrell> busje komt zo :p
[14:21:53] <Maas> Better add the protocol too I guess.
[14:22:37] <Marc Worrell> you used "//foo.bar" ?
[14:22:52] <Maas> Not even sure what I used :-)
[14:23:42] <Maas> window.location.protocol + "//" ...
[14:23:57] <Maas> maybe the protocol was undefined. weird
[14:26:23] <Maas> Yup, doesn't work.
[14:26:42] <Arjan> isn't it document.location.protocol
[14:26:46] <Arjan> isntead of window?
[14:28:18] <Maas> it is the same location thing, and I now see it does work, only console.log works a bit different on android
[14:43:04] <Maas> Oops, typo somewhere, other browsers don't seem to be bothered. Luckily I can fix my own bugs.
[16:11:34] <Andreas Stenius> Arjan: I esl called me, and while I had them on the phone asked them about the talk submissions. They're going through them all this week, so you should hear from them soon...
[16:15:36] <Arjan> cool! thanks :)
[17:24:40] Maas leaves the room
[17:50:10] Arjan leaves the room
[18:04:10] Andreas Stenius leaves the room
[18:08:29] Andreas Stenius joins the room
[19:50:41] Arjan joins the room
[19:52:09] Arjan leaves the room
[21:10:08] arcusfelis leaves the room
[21:11:38] Arjan joins the room
[21:48:50] Arjan leaves the room
[22:13:35] Marc Worrell leaves the room: Disconnected: connection closed
[22:13:39] Marc Worrell joins the room
[22:35:46] Andreas Stenius leaves the room
Powered by ejabberd Powered by Erlang Valid XHTML 1.0 Transitional Valid CSS!