<havingtrouble>
I replaced the domain with XXX, obviously
<havingtrouble>
during the install I chose 'none' to skip the "sandcats" stuff
<havingtrouble>
oh... the port number on that BASE_URL is a typo when I changed the domain in the pastebin. it is 6080 in the conf file
jemc has quit [Quit: WeeChat 1.4]
<havingtrouble>
see anything off?
jemc has joined #sandstorm
<asheesh>
havingtrouble: Hi! Let me look again now; sorry had to afk
<asheesh>
BASE_URL needs to say http:// at the start
<asheesh>
And WILDCARD_HOST needs to say *.xxx.xxx.xxx:6080
<havingtrouble>
ah.
<asheesh>
I should fix the error log message to indicate this problem more clearly. I'll file a bug once you indicate that we've solved your problem, havingtrouble
<asheesh>
BTW mega thanks for coming by and getting help.
<havingtrouble>
:D Mega thanks for taking some time to help me!
<havingtrouble>
I'll make those changes and see what happens
<asheesh>
Great! Keep me posted.
<havingtrouble>
that was it!
mnutt_ has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
<havingtrouble>
^5 awesome!
<asheesh>
BTW consider using sandcats.io for dyndns so you don't have these problems. (-:
<havingtrouble>
:D will sandcats domain work when the server is in a private vpc?
<dwrensha>
it is the app's responsibility to handle migration of data
<dwrensha>
one thing you could do: maintain a /var/migrations directory with an empty file for each migration that has already taken place
<dwrensha>
when the app boots, it can check whether all of its needed migrations are done, and if not, it can perform the remaining ones
<dwrensha>
often apps already have built-in migration handling. For example, WordPress remembers its schema version, and automatically migrates when necessary
<dwrensha>
On the other hand, updating nginx or php typically shouldn't require a migration
<dwrensha>
an spk is a self-contained set of files required to run the app. When you publish a new spk, you provide an entire new set of files.
<dwrensha>
xming: does that answer your question?
amyers has joined #sandstorm
<xming>
dwrensha: yes thanks, gives me an idea
<xming>
so who's responsibility it is for checking /var/migrations? My own I suppose
<xming>
And more questions, does sandstorm run on ARM? and does it run inside s lxc container?
<dwrensha>
currently only runs on x86_64, mainly because we don't want to fracture the app ecosystem
<dwrensha>
I think you should be able to run Sandstorm in an lxc container
<xming>
Ideally I want to run it inside lxc on ARM
<xming>
dwrensha: even if I compile and manage sandstorm from source, I can't get it to work on ARM?
<xming>
going to see if it compiles on ARM
<dwrensha>
I think that mongodb does not support ARM
<xming>
to make it even weirder, musl libc on ARM
<dwrensha>
someone had I link to an issue tracker page about mongo on ARM, but I'm having trouble finding it at the moment
<dwrensha>
Sandstorm depends on Mongo, so getting Mongo to compile is necessary
<xming>
np, I've some while googling
<xming>
by they don't really support 32-bit, some brave souls have mongodb running on armv7 (32-bit) so this is encouraging
<xming>
the question is do I want this? :/
<dwrensha>
note that none of the current app packages will work on ARM
<xming>
yes I am fully aware of that
<xming>
unless it's hooked with ExaGear :D
larjona has quit [Remote host closed the connection]
frigginglorious has joined #sandstorm
larjona has joined #sandstorm
rolig has quit [Ping timeout: 264 seconds]
rolig has joined #sandstorm
amyers has quit [Remote host closed the connection]
amyers has joined #sandstorm
keybits has joined #sandstorm
NOTevil has joined #sandstorm
NOTevil has left #sandstorm [#sandstorm]
frigginglorious has quit [Quit: frigginglorious]
jemc has quit [Quit: WeeChat 1.4]
frigginglorious has joined #sandstorm
jemc has joined #sandstorm
jemc has quit [Quit: WeeChat 1.4]
jemc has joined #sandstorm
raoulzecat has joined #sandstorm
frigginglorious has quit [Quit: frigginglorious]
raoulzecat has quit [Remote host closed the connection]
raoulzecat has joined #sandstorm
frigginglorious has joined #sandstorm
mnutt_ has quit [Quit: My Mac has gone to sleep. ZZZzzz…]
prettyvanilla has quit [Quit: Konversation terminated!]
prettyvanilla has joined #sandstorm
prettyvanilla has quit [Excess Flood]
prettyvanilla has joined #sandstorm
amyers has quit [Ping timeout: 244 seconds]
kecors has quit [Quit: Leaving]
amyers has joined #sandstorm
amyers has quit [Ping timeout: 246 seconds]
<Daemoen>
whats up folks
<neynah>
Hi!
<Daemoen>
how are you neynah :)
<xming>
grrr meteor is problematic on unsupported platform, anyway to bypass that?
<neynah>
I'm having a good day Daemoen. How about you? :D
<Daemoen>
trying to have a productive day but getting very angry. this is my second week at $curr_contract and I still am blocked on access and basically just sitting at my desk doing absolutely nothing, which drives me *insane*, because I get judged for not doing anything, but I am blocked on access...
<neynah>
I'm sorry to hear that. That sounds pretty frustrating. :(
<Daemoen>
yepyep
<Daemoen>
its ok though, maybe ill spend some time working on xwiki+sandstorm, i just dont want to work on something personal while being paid under contract :\
frankier has quit [Ping timeout: 240 seconds]
<asheesh>
xming: Hypothetically I guess you could do a 'meteor build' on a supported platform, then look at the artifacts it found, and swap the architecture-dependent ones for ones corresponding to a different architecture.
<asheesh>
That might work pretty well, actually, given that you're supposed to then 'npm install' *after* 'meteor build', so you could do the 'npm install' phase on your actual target architecture.
<xming>
does sandstorm needs nodejs at runtime?
<asheesh>
Yeah, for the main web UI.
<xming>
so theoretically the target doesn't need any meteor bits, just npm install the missing node deps?
<asheesh>
I *think* "meteor build"'s job is to download meteor-related code so that the Meteor command line tool is not needed at runtime, which I think is a yes to your question.
frigginglorious has quit [Quit: frigginglorious]
frigginglorious has joined #sandstorm
<xming>
okay thanks I will have to learn the pieces of the build process
<xming>
does sandstorm create a container for each app with its own IP?
<asheesh>
Yes a unique network namespace, but it only has loopback connectivity.
<xming>
awe, only http(s)? no other UDP or TCP connections allowed to the outside world?
<Daemoen>
asheesh: looked over the documentation and there is one thing that im surprised hasnt been posted anywhere, unless im just blind... How is HA/clustering handled within sandstorm ?
<xming>
so it's lny for webapps
<xming>
only*
<zarvox>
No direct connections out of the box, but you can request IpNetwork/IpInterface capnproto capabilities through the powerbox, which allow for direct network access.
<zarvox>
Pieces of the powerbox, including the ability to request and provide IpNetwork and IpInterface, are available today.
<xming>
so normal iptables for NAT/PAT
gkoz has quit [Read error: Connection reset by peer]
<zarvox>
We have not yet implemented selection UI for application-defined powerbox interfaces.
<xming>
but the apps can make requests albeit w/o user knowledge, is t hat correct?
<zarvox>
yeah, configuring your host's firewall rules is something beyond Sandstorm's scope
<xming>
zarvox: yeah I suspected that
<zarvox>
requests are only fulfilled via user action. In particular, only server admins can create an IpNetwork or IpInterface and pass it to a requesting grain.
<xming>
Oh
<zarvox>
The apps can request a pony, but until a user grants them one, they have nothing :P
<xming>
w/o web UI, are there cli tools for that?
joshbuddy has joined #sandstorm
<Daemoen>
asheesh: so does that in turn mean that all clustering and HA is going to be only available in blackrock until the community shows enough interest for it to become part of sandstorm proper? (thats how it reads, so just clarifying)
<zarvox>
xming: You'd need a web-ui piece to make the request to the user. You'd probably want a web-ui piece to show the grain owner useful contextual information.
<asheesh>
Daemoen: Yeah, that is the plan for now. Note also that we're not really spending much time working on that, just barely enough to run the hosting service that we're now running (oasis.sandstorm.io).
<asheesh>
For scaling, there are APIs within Sandstorm itself that enable something else to scale it, and those APIs are what our other code uses.
<Daemoen>
asheesh: gotcha. from a technical perspective, it will be a complex undertaking, since you will have to decide on which approach makes more sense... do you handle clustering at the sandstorm level, or at the grain level... take for instance rocketchat... sandstorm primary server with rocketchat dies... but there is a secondary server running same... is it at the grain or 'top' level =D
<asheesh>
The other thing to consider is auto-scale-down within Sandstorm -- this already works and is an important feature, since otherwise grains would run forever and consume too much RAM to be reasonable.
<asheesh>
Well you have to scale both, Daemoen (-:
<Daemoen>
asheesh: right... but... my point is, do you allow sandstorm to manage it all at the top level, or do you enable each grain individually to manage it, or both, or a mix of the above options -- each of these scenarios brings its own difficulties and complexities. sorry, dont want to spam the channel, just something I had noticed wasnt covered much on any of the docs i had seen =D
raoulzecat has quit [Ping timeout: 268 seconds]
keybits has quit [Ping timeout: 244 seconds]
<zarvox>
The general design is that grains run on a single node, and blackrock is responsible for scheduling those such that any given worker machine can die and the service remains up.
jadewang has quit [Remote host closed the connection]