<digitalcircuit>
cstrahan: I'm looking forward to an IRC Sandstorm app, too. In the meantime, Quassel works well to keep data under your control, though set-up takes more. I think someone mentioned considering porting Quassel-web to Sandstorm.
<zarvox>
If you do give it a go, let me know what works, where things broke down, and if you have any other issues :)
<zarvox>
I'll probably throw it in the vagrant-spk repo or wiki or something
<chilts>
thanks
<chilts>
first problem on Ubuntu 14.04 : E: Unable to locate package libvirtd
<chilts>
is that in a particular package repo perhaps?
<chilts>
didn't think I could see it earlier either, just libvirt0
<chilts>
and libvirt-bin
|jemc| has joined #sandstorm
<zarvox>
oh, yeah, probably libvirt-bin (which will pull in libvirt0)
<zarvox>
and libvirt-daemon and libvirt-daemon-system, probably?
<zarvox>
libvirt-clients is probably what provides virt-manager
bb010g has quit [Quit: Connection closed for inactivity]
<zarvox>
oh, there's a quality-of-life thing I forgot about involving writing a polkit rule to allow your user to do libvirt calls without having to reauthorize all the time
bb010g has joined #sandstorm
relicanth has joined #sandstorm
relicanth has quit [Client Quit]
relicanth has joined #sandstorm
relicanth has quit [Client Quit]
frigginglorious has quit [Quit: frigginglorious]
wolcen has joined #sandstorm
* asheesh
waves to all
* phildini
waves back
dvn has joined #sandstorm
frigginglorious has joined #sandstorm
rustyrazorblade has joined #sandstorm
rustyrazorblade has quit [Quit: rustyrazorblade]
|jemc| has quit [Ping timeout: 240 seconds]
rustyrazorblade has joined #sandstorm
rustyrazorblade has quit [Quit: rustyrazorblade]
kentonv has quit [Quit: Leaving]
frigginglorious has quit [Quit: frigginglorious]
raoulzecat has joined #sandstorm
jadewang has quit [Remote host closed the connection]
jadewang has joined #sandstorm
jadewang has quit [Remote host closed the connection]
raoulzecat has quit [Ping timeout: 255 seconds]
raoulzecat has joined #sandstorm
jadewang has joined #sandstorm
jadewang has quit [Ping timeout: 240 seconds]
<zarvox>
blah, I need ed25519 bindings for node, which will probably involve native code
axx_ has joined #sandstorm
mrshu has quit [Ping timeout: 260 seconds]
mrshu has joined #sandstorm
jadewang has joined #sandstorm
synchrone has joined #sandstorm
indiebio has quit [Remote host closed the connection]
jadewang has quit [Ping timeout: 250 seconds]
<zarvox>
looks like native modules pulled from npm Just Work, so that's nice
synchrone has quit [Ping timeout: 250 seconds]
mrshu has quit [Ping timeout: 244 seconds]
mrshu has joined #sandstorm
axx_ has quit [Ping timeout: 255 seconds]
jadewang has joined #sandstorm
axx_ has joined #sandstorm
jadewang has quit [Ping timeout: 276 seconds]
synchrone has joined #sandstorm
axx_ has quit [Ping timeout: 250 seconds]
NwS has joined #sandstorm
jadewang has joined #sandstorm
jadewang has quit [Ping timeout: 244 seconds]
TwoJames has quit [Ping timeout: 268 seconds]
jadewang has joined #sandstorm
TwoJames has joined #sandstorm
jadewang has quit [Ping timeout: 240 seconds]
synchrone has quit [Remote host closed the connection]
synchrone has joined #sandstorm
axx_1 has joined #sandstorm
amyers has joined #sandstorm
TwoJames has quit [Ping timeout: 248 seconds]
jadewang has joined #sandstorm
jadewang has quit [Ping timeout: 240 seconds]
axx_1 has quit [Ping timeout: 255 seconds]
ncls has joined #sandstorm
<ncls>
hey
jadewang has joined #sandstorm
axx_1 has joined #sandstorm
jadewang has quit [Ping timeout: 240 seconds]
frigginglorious has joined #sandstorm
axx_1 has quit [Ping timeout: 240 seconds]
synchrone has quit [Ping timeout: 240 seconds]
joshbuddy has quit [Quit: joshbuddy]
axx_1 has joined #sandstorm
<asheesh>
Hey ncls !
<ncls>
Tried to install Sandstorm on a dedicated server but I ran into the "Your kernel does not appear to be compiled with support for unprivileged user" error
<asheesh>
ncls: Ah, gotcha
<ncls>
as I'm not a sysadmin, I don't know if any easy work around ..
<asheesh>
What Linux distro are you using?
<ncls>
exists
<asheesh>
You might have to switch what Linux kernel is used.
<asheesh>
I can possibly help you with that.
<asheesh>
What dedicated server provider, if you don't mind letting me know?
<ncls>
Ubuntu 15.10 but I think it's a custom kernel from the provided
<asheesh>
Also, I hope you'll consider using a "long term support" version of Ubuntu, such as 14.04, so that you get security updates from Ubuntu for a little longer.
<asheesh>
It's not a huge deal since Sandstorm has its own security hardening, but it might give you even more peace of mind.
<asheesh>
Anyway ncls long story short, you'll have to switch kernel; let me know if you can try following one of those guides!
<ncls>
thanks asheesh ! I'll take a look at this, and if it doesn't seem too complicated, I'll do it !
<asheesh>
OK, great!
<ncls>
looks pretty easy in the first link, but I have no idea what this implies
<ncls>
other question, is it difficult to configure sandstorm in order to set it up "behind" an Apache server ?
<asheesh>
It's easiest if it has its own server, but it's not crazy hard to make work the other way.
<asheesh>
It should work fine behind Apache, with one caveat that Sandstorm's built-in free HTTPS certificates are something it needs to control, so you need sniproxy rather than Apache/nginx if you want to use its built-in free SSL.
<asheesh>
There's docs on that, so it should be pretty doable.
<phildini>
i,i "sandstorm hosted graphite and grafana"
<chilts>
someone was talking about being able to spin up some kind of log server yesterday
<chilts>
perhaps that's another one to add to that collection :)
<phildini>
chilts: I think that was me again? I just really, really want it. :)
xet7 has joined #sandstorm
<zarvox>
:)
<chilts>
phildini: hah, awesome :)
<chilts>
I need to switch off joins and parts in my irc client
<phildini>
I'm fairly curious about how piwik works, since isn't "always on" against the sandstorm container ethos?
<chilts>
I started on doing a .sandstorm/build.sh file yesterday - that's my latest status
<chilts>
I think it'll take me a while to get that and all the other .sandstorm files done, though I hope I'm picking it up quickly enough
<zarvox>
phildini: the piwik grain gets launched when a request with the appropriate Authorization: header is sent to the api endpoint
<zarvox>
it services the request, and then if nobody visits the page for a while, then it'll shut down again
<chilts>
so if it constantly gets request, it'll just keep the grain alive?
<chilts>
*requests
<phildini>
hmm... two things come to mind: 1) I hope that launch is fast? 2) with a high-enough traffic site, the grain will never go down.
<zarvox>
Yeah. It may get killed and relocated at some point. But yes, we aim for launch to be fast, and the request will be held at the proxy until the backend is up, so you shouldn't drop requests.
<chilts>
nice
funwhilelost has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<chilts>
just wondering if there is an indication on (a) how long after the last request the grain might get killed, or (b) how often an always up grain might get re-allocated?
<chilts>
In Heroku they talk about around once a day for (b)
<chilts>
and I think 30mins for (a)
<chilts>
(for dynos that an sleep)
<chilts>
s/an/can/
<zarvox>
"implementation details" but I think the timeout for (a) is something like 3 minutes (and the shell sends keepalives every minute, if you have the grain open in the Sandstorm UI)
<chilts>
I like the idea of fast-launching grains, since I think Heroku's 30s or so startup time is a service killer (ie. not great)
<zarvox>
and I can't say anything in particular about (b) - today, it should be rare that the grain gets moved around, but in the future we want to be able to relocate grains for load balancing/rolling upgrade reasons
<zarvox>
Yeah, 30s is basically unacceptable for user-facing things.
<zarvox>
Piwik is at the high end of what I can tolerate. It probably has room for optimization, particularly around skipping migrations if there hasn't been a package upgrade since the last one completed successfully.
kecors has quit [Quit: Leaving]
<zarvox>
phildini: what all goes into a graphite deployment? there's the daemon that runs on your servers, your aggregator, and the thing that displays the graphs?
<phildini>
yeah... normally the aggregator and the display-er are the same server.
<zarvox>
okay, so maybe those two pieces would live in a grain
<chilts>
zarvox: I'm looking at also adding database migrations into my (as yet unreleased) app, so I've been wondering how an app upgrade works/happens. Is there a doc I can read about that? I'm pretty alright on how to deal with DB migrations, but just wondering how it works in Sandstorm. I think I found a helpful thing yesterday related to /var being stable across grains launching, so perhaps that's my answer (put the database in there). :)
<zarvox>
and how do you normally get data into the aggregator?
<chilts>
(sorry for the long question)
<zarvox>
chilts: yeah, put your DB in /var. Then, you have to make sure that future releases of your app can handle all of the possible /var's created by previous releases of your app.
<chilts>
sweet, that's it then
<chilts>
and also file uploads in there too
<zarvox>
And then you have to run any DB migrations on grain launch, in case your project changed schemas between releases
<chilts>
yeah, that's my plan
<chilts>
I guess I could do that in a .sandstorm file, or inside the app itself
amyers has quit [Ping timeout: 276 seconds]
<zarvox>
yeah, either .sandstorm/launcher.sh or "app takes care of itself" are good approaches
* chilts
goes back to read where and when each script runs
<asheesh>
puck2: If 443 and 80 are avail, and you choose sandcats, it uses 443+80 instead.
<asheesh>
Otherwise, that's right.
<asheesh>
Also, morning all.
<asheesh>
mrdomino: I wonder what "netflix socks" is; maybe I can Google it.
<asheesh>
https://makeit.netflix.com/ seems like the kind of thing that'd be an April Fools joke! But I guess it's real!
<puck2>
thanks, asheesh. good morning to you
<asheesh>
phildini + chilts: I also want these log storage apps etc.
joshbuddy has quit [Quit: joshbuddy]
notevil has quit [Quit: Leaving]
<mrdomino>
dwrensha: chromium on os x. it appears to be getting reset, but i'm not sure. i get stream interruptions every now and then. when i look in the log, i see "Grain no longer in use" immediately followed by the startup and shm_open fail
amyers has quit [Ping timeout: 244 seconds]
<dwrensha>
I think the shm_open fail is a red herring
<mrdomino>
yeah
<dwrensha>
I've also seen such behavior
<dwrensha>
the client is supposed to send a keepalive every minute
<dwrensha>
and the sandstorm supervisor shuts down the grain if it hasn't seen a keepalive within 90 seconds
<mrdomino>
so either there's 30s+ of time in which a keepalive isn't going through, or it's something else shutting down the grain?
<dwrensha>
seems likely that somehow the keepalive isn't getting through. I'm not sure where the problem is, though.
<dwrensha>
one hypothesis is that the browser is being lazy about executing the setInterval() callback
<mrdomino>
yeah, if the tab's in the background for a while i could see that happening
<dwrensha>
however, I seem to remember observing the problem even when the tab is in the foreground
<dwrensha>
I'll try inserting some tracing into that setInterval() on my server...