Switch to DuckDuckGo Search
   October 17, 2013  
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | >

bottom
[00:10:11] *** pringlescan has quit IRC
[00:29:26] *** jim80net has quit IRC
[00:57:32] <trentster> if pringlescan comes back his VNC issue is very likely related a change introduced in Chrome 30 that breaks websockets/novnc there is a ticket logged for this to be fixed in the fifo code. In the interim, using an earlier version of Chrome or alternative browser will work.
[00:58:30] <trentster> nahamu: I have updated the "requirements page" and included some additional info there as well. ;-)
[01:05:21] <arekinath> alternatively, just comment out the protocol parameters to websockets, also a temporary workaround. :P
[01:05:37] <arekinath> and one that doesn't require downgrading chrome, which is surprisingly hard
[01:05:57] <trentster> arekinath: gist the change and file name please
[01:10:13] <arekinath> it's a few files and I forgot to keep old copies when I went hackity hack
[01:10:16] <arekinath> might be a while
[01:10:17] <arekinath> haha
[01:10:52] <arekinath> just find anywhere someone does new Websocket(..., ...) (the two-parameter version)
[01:10:58] <arekinath> and comment out the second parameter
[01:14:49] <trentster> erm, sounds very hackish to attempt, especially when not sure which code to change where :-P
[01:15:18] <trentster> I will await proper instructions, or until Licenser closes the ticket.
[01:24:21] <arekinath> trentster: link to ticket?
[01:29:47] <arekinath> WIG-43?
[01:31:07] <trentster> yup wig-43
[01:31:29] <trentster> arekinath: correct http://jira.project-fifo.net/browse/WIG-43
[01:31:32] <trentster> thanks mate
[01:39:39] *** noahmehl has quit IRC
[01:41:14] *** noahmehl has joined #project-FiFo
[01:57:53] <trentster> arekinath: thanks, read your ticket comment, so if I understand correctly what you are saying, is that fifo needs to adhere to RFC in order to work correctly with future browsers. "Cowboy" does support the header stuff correctly, but changes have to be made to wiggle to pull out the headers correctly?
[01:58:29] <trentster> or do we have to wait for an essen to update cowboy to support it better?
[01:58:50] <arekinath> trentster: well, cowboy will probably change eventually to support websocket subprotocols better than it does now
[01:59:01] <arekinath> trentster: but I don't think we should wait, it could be a while by the look of it
[01:59:16] <arekinath> trentster: so we can add a bit of code to wiggle and howl to deal with it for now
[01:59:36] <arekinath> I followed up with a gist too, with a vague idea of how to do it for wiggle at least
[02:00:26] <trentster> hmm, so is it something perhaps you or Licenser can fix in the code that does not mean disabling the RFC checks?
[02:00:55] <trentster> is that what you mean by "I dont think we should wait" ?
[02:01:26] <arekinath> if you apply that gist, it should work again in wiggle
[02:01:30] <arekinath> similar fixes will work for howl too
[02:02:07] <arekinath> when cowboy changes what they're doing we might want to revisit it, that's all
[02:02:20] <trentster> yeah, I got that, I am more concerned with having it work out the box for our users?
[02:03:23] <arekinath> what do you mean exactly? that patch will fix up the problem, but it'll have to be in the next release
[02:04:10] <trentster> or else I must put a big warning on the wiki areas either denoting supported browser versions, or a hack/patch guide to fix if they are using chrome 30.
[02:05:01] <arekinath> maybe Licenser can spin a maintenance release with just that patch (after we finalise it) and we can point people at upgrading to that
[02:05:38] <trentster> thanks, back in a bit
[02:05:38] <arekinath> since it'd just be a pkgin upgrade and svcadm restart away
[02:51:34] <killfill> hi
[04:09:50] *** pringlescan has joined #project-FiFo
[04:40:31] *** pringlescan1 has joined #project-FiFo
[04:41:25] *** pringlescan has quit IRC
[05:49:07] <trentster> ok back.
[05:49:12] <trentster> howdy killfill
[05:50:56] *** pringlescan1 has quit IRC
[05:51:09] *** pringlescan has joined #project-FiFo
[05:54:25] *** pringlescan has quit IRC
[05:54:37] *** pringlescan has joined #project-FiFo
[05:57:02] *** pringlescan1 has joined #project-FiFo
[05:59:19] *** pringlescan has quit IRC
[06:01:33] *** pringlescan1 has quit IRC
[06:32:14] *** noahmehl_ has joined #project-FiFo
[06:33:31] *** noahmehl has quit IRC
[06:33:32] *** noahmehl_ is now known as noahmehl
[08:40:32] *** noahmehl has quit IRC
[09:15:55] *** wiedi has joined #project-FiFo
[09:17:42] *** alcir has joined #project-FiFo
[09:31:04] *** mgls has joined #project-FiFo
[10:33:00] <Licenser> aloa
[10:34:28] <blindcoder> morning
[10:35:42] <MerlinDMC> hello
[10:45:27] *** alcir has quit IRC
[11:15:17] <trentster> greetings
[11:25:06] *** alcir has joined #project-FiFo
[11:25:16] <alcir> so
[11:25:32] <alcir> I wanto to see the hostname instead of the uuid of the GZs
[11:25:38] <alcir> there is a way?
[11:25:53] <alcir> can I change /opt/chunter/etc/host_id?
[11:25:53] <trentster> aah no alcir you broke the "greeting" chain ;-)
[11:26:08] <alcir> sorry
[11:26:12] <alcir> buongiorno
[11:28:19] <trentster> alcir: yes, but I think Licenser mentioned its best practice to leave it alone.
[13:49:27] <Licenser> alcir it is
[13:51:16] <trentster> The master has arisen :-P
[13:52:12] <alcir> hi master
[13:52:26] <alcir> but I can change it when I want?
[13:54:25] <trentster> Licenser: aka the master - ping
[13:54:35] <Licenser> alcir yea but it voids warrenty
[14:11:11] *** _lb_ has joined #project-FiFo
[15:17:12] *** wiedi_ has joined #project-FiFo
[15:17:54] <nahamu> trentster: thank you!
[15:19:22] <trentster> nahamu: for?
[15:19:49] <nahamu> updating the wiki page.
[15:20:16] *** wiedi has quit IRC
[15:20:57] <trentster> aah, no problemmo ;-)
[16:09:33] *** jim80net has joined #project-FiFo
[16:26:49] *** noahmehl has joined #project-FiFo
[16:30:52] *** pringlescan has joined #project-FiFo
[16:31:01] <pringlescan> hello all
[16:31:07] <pringlescan> Licenser: how's it going :-)?
[16:32:15] <Licenser> good good :) fighting chrome at the moment
[16:33:47] <pringlescan> Oh! I'm good with that.
[16:33:58] <pringlescan> I was having some web socket issues proxying nginx through nginx with Project-FIFO.
[16:34:24] <Licenser> newest jingles/wiggle also solve it
[16:34:38] <pringlescan> But more so, I'm trying to reinstall chunter (rel) on a hypervisor it was already on, and even when I delete the service and remove the directory and run the install script, it's unable to start and goes into maintenance mode without much detail in the error log
[16:34:42] <pringlescan> is there something else I can try?
[16:35:18] <pringlescan> I'm leaving the job where I set up 4 smartos servers and I always used the command line, but it's just IT staff who will be manning it and they want a GUI, and FIFO is going to be a lifesaver for them
[16:35:24] <pringlescan> and it's so damn pretty
[16:35:36] <pringlescan> they kept saying "and this is open source? but it's not …. ugly"
[16:35:42] <pringlescan> so wanted to relay their complements
[16:36:01] <Licenser> pringlescan I would recommand using dev ;)
[16:36:13] <pringlescan> and the installers will remain as easy as the new one?
[16:36:32] <Licenser> that's the plan
[16:36:35] <pringlescan> I don't want to leave something that they need to frequently update, but if I can leave them easy instructions
[16:36:42] <pringlescan> I was impressed it works really well, it's hard to write good installers
[16:37:01] <Licenser> well it's pkgin that's doing all the workd ;)
[16:37:08] <Licenser> but thanks ^^ much appriciated
[16:37:53] <pringlescan>  now, I could look at the code, but it looks for a particular image and if it's not there it uses another one rather than trying to retrieve it
[16:37:55] <pringlescan> is that desired?
[16:39:14] <Licenser> where exactly?
[16:40:00] <pringlescan> Could not find image 286f8c9a-24d9-4f85-88f6-aec98cf3a1b4 in image sources
[16:40:00] <pringlescan> Image fdea06b0-3f24-11e2-ac50-0b645575ce9d (base64 1.8.4) is already installed, skipping
[16:40:04] <pringlescan> when running the installer
[16:40:24] <pringlescan> also if the fifo zone can't get to the internet the installer just hangs but I wouldn't necessary plan for that
[16:40:42] <Licenser> ah the one step installer
[16:40:46] <pringlescan> I manage to break everything you must understand :-)
[16:40:57] <Licenser> that's fine :P
[16:41:16] <Licenser> if the zone can't get to the internet you can't install ;)
[16:41:18] <pringlescan> now for auto updates of dev, what's the easiest way
[16:41:56] <pringlescan> I was thinking of doing a cluster fifo deployment but I didn't know if communications were safe over the public internet or if I'd need a VPN
[16:42:34] *** noahmehl_ has joined #project-FiFo
[16:42:39] <Licenser> it needs a l2 connection so it won't work over the internet
[16:42:39] <pringlescan> ah you have diagrams
[16:42:57] <Licenser> http://project-fifo.net/display/PF/Upgrade+Guide#UpgradeGuide-0.4.8
[16:43:03] <Licenser> for updating rel->dev
[16:43:23] <Licenser> and just install the latest chunter
[16:43:36] <pringlescan> so you'd do that in each fifo zone and there will be instructions there for each version?
[16:44:19] *** noahmehl has quit IRC
[16:44:20] *** noahmehl_ is now known as noahmehl
[16:45:44] <Licenser> There is a general update guide at the top, which usually applies
[16:45:55] <Licenser> that is just update the package
[16:46:14] <Licenser> going from rel->dev or dev->rel also requires changing the repositiry
[16:46:15] <pringlescan> when I tunnel 443 and 80 the fifo instance, I get WebSocket handshake: Sec-WebSocket-Protocol mismatch's in Chrome with a dev version of fifo
[16:46:58] <Licenser> but some versions require special extra steps, if there are extra steps required / then they're noted for that version (as 0.4.8)
[16:47:19] <Licenser> 0.4.2 darn
[16:47:39] <pringlescan> weird
[16:47:57] <pringlescan> So the problem I have is "Node chunter@sts1 not responding to pings."
[16:48:50] <pringlescan>  /opt/chunter/bin/update -> no update
[16:48:59] <pringlescan> I try removing it and installing it from scratch
[16:49:15] <Licenser> install dev ;)
[16:49:22] <pringlescan> I thought I did :-)
[16:49:34] <pringlescan> yes, I did chunter-latest?
[16:49:40] <pringlescan> dev/chunter-latest rather
[16:50:27] <pringlescan> I removed /opt/chunter and /opt/chunter-latest and I will trya gain
[16:50:56] <pringlescan> ah, it's staying online now
[16:51:57] <pringlescan> fifo can't see it but I'll restart both and hopefully they'll talk. I did a fresh install of release 0 issues.
[16:52:04] <pringlescan> [on a machine that never had fifo before]
[16:52:34] <Licenser> have you looked at the troubleshooting guide?
[16:53:00] <pringlescan> Not yet, I have to see how the IT staff will handle this
[16:53:07] <pringlescan> They don't read documentation very well
[16:53:46] <pringlescan> I search for troubleshooting and trouble in the wiki and get nothing, weird
[16:54:26] <Licenser> if they don't read documentation very well fire them :P
[16:54:33] <pringlescan> I quit :-)
[16:54:39] <pringlescan> giving them 30 days to get them up to speed
[16:54:51] <Licenser> also please don't install fifo then
[16:55:26] <pringlescan> The guy who will be using FIFO is pretty god. He downloaded SmartOS to vmware on his own and started playing with it before he met with me on how to use it
[16:55:34] <Licenser> I'm not going to put up with helping people who refuse to read the documentation
[16:55:41] <pringlescan> so he wants to learn, I have faith in him :-) but he's not as good as finding help
[16:56:06] <pringlescan> how does this sound? it's from my "SmartOS -> How to get help"
[16:56:32] <pringlescan> Use SmartOS search. Search the IRC logs, Search the mailing list archive. Do not ask for help in IRC or the mailing list without doing the prior 3.
[16:56:42] <Licenser> heh it sounds sad :P
[16:57:02] <pringlescan> well, he's never been on IRC before, I was a wee baby when I got beat up on IRC for being a "n00b"
[16:57:15] <pringlescan> it also warns him that everything is public record in these mediums
[16:57:19] <pringlescan> it's pretty funny actually
[16:57:46] <pringlescan> http://project-fifo.net/display/PF/Problem+Checklist is this what you meant by troubleshooting guide?
[16:57:53] <Licenser> @websocket newst dev works with newest chrome
[16:58:00] *** wiedi_ has quit IRC
[16:58:06] <Licenser> yup it has instructions on how to ensure the netwokring works
[16:58:18] <Licenser> since if chunter/sniffle don't see eachother that is usually the underlaying problem
[16:58:25] <pringlescan> ah chunter fell back into maintenance
[16:58:46] <Licenser> can you tar up the logs and send them over?
[16:59:57] <pringlescan> it's pretty full in there, some might be from prior releases, should I clear it out, reproduce just my new problem and then send it over, or would you like the whole shebang?
[17:00:15] <Licenser> cleaned up is much easyer :)
[17:00:55] <pringlescan> now when I do a svcadm clear chunter, it doesn't even get far enough to create logs
[17:01:06] <Licenser> chrome 30 changed the way it handles the sec-protocol stuff to a more strict model which was too strict for the way fifo handled it <- that's why you had the ussues
[17:01:07] <pringlescan> oh, maybe the logs need to be touched first?
[17:01:16] <Licenser> not if it's stopped
[17:01:46] <pringlescan> I will try to reinstall again
[17:01:49] <pringlescan> I got it to run briefly
[17:02:08] <Licenser> could it be that there is some kind of leftover server running?
[17:02:36] <pringlescan> a bunch of these are leftover: root     20156  0.0  0.0 5744 2288 ?        S 21:43:19  0:00 /opt/chunter/lib/chunter-0.3.11/priv/zonedoor 2854ebe9-ed87-45f8-8d91-dcce788c7bc9 _joyent_sshd_key
[17:02:45] <pringlescan> restarting the erlang daemon will fix that right?
[17:03:09] <Licenser> ah that is a sucky problem :(
[17:03:12] <Licenser> best kill them
[17:04:16] <pringlescan> ps -ef | grep chunter | grep -v grep | awk '{print $2}' | xargs kill -9
[17:04:18] <pringlescan> that worked
[17:05:05] <pringlescan> so far the zone hasn't found it yet
[17:05:09] <pringlescan> *fifo zone
[17:05:39] <pringlescan> now back to the troubleshooting guide
[17:05:48] <pringlescan> maybe we should add that to the chunter installer?
[17:06:22] <nahamu> pringlescan: "pkill" perhaps?
[17:07:08] <pringlescan> if you want to be concise, I come from a no-kill shop
[17:07:57] <pringlescan> [root@ecdf0cf1-3673-4df2-9e86-a97e120fbb3b ~]# ping ping project-fifo.net
[17:07:57] <pringlescan> ping: bad timeout: project-fifo.net
[17:08:05] <pringlescan> but I can ping 8.8.8.8
[17:08:14] <pringlescan> and google.com resolves
[17:08:21] <pringlescan> ah ping ping :-)
[17:08:55] <pringlescan> hm, everything checked out in the problem checklist
[17:10:27] <pringlescan> Licenser: 2013-10-17 15:04:15.244 [warning] <0.220.0>@chunter_vm_fsm:handle_info:512 unknown data: {#Port<0.3159>,{exit_status,1}} is that important?
[17:10:54] <pringlescan> there's really nothing else going on in the chunter logs, it opened up it's zone doors
[17:11:27] <Licenser> that might be a issue
[17:12:10] <pringlescan> I'm trying to reboot the fifo zone, anything else I should try? I could send over the logs, but I'm fairly sure there's nothing notable
[17:12:41] <pringlescan> I did svcadm restart epmd, then svcadm restart chunter, and it went into maintenance
[17:12:45] <Licenser> if that's the only thing no need for the logs
[17:13:05] <Licenser> and yea it mens one of the zones is acting up
[17:13:17] <pringlescan> Now I get a crash report
[17:13:28] <pringlescan> 2013-10-17 15:11:50.056 [error] <0.246.0> gen_server chunter_perf_plugin terminated with reason: bad argument in call to erlang:list_to_binary(undefined) in zone_cpu:build_obj/2 line 21
[17:13:42] <pringlescan> so if you start chunter while a zone is rebooting, that will happen
[17:13:44] <pringlescan> I can try to reproduce
[17:13:48] <pringlescan> but that seems to be the issue
[17:14:02] <Licenser> pringlescan that sounds like something being wrong with your smartos
[17:14:14] <pringlescan> hm, no it still goes into maintenance mode
[17:14:32] <pringlescan> what version are you running? I think I told them to stay back a version and watch the change log
[17:15:04] <pringlescan> o_O platform-upgrade "Checking current boot device... detected mount: No such file or directory
[17:15:04] <pringlescan> , mount failed"
[17:15:06] <pringlescan> that's not good
[17:15:33] <pringlescan> It's running: 20130822T211238Z
[17:16:40] <pringlescan> I'm afraid to reboot that box unless i'm sitting in the datacenter to troubleshoot the USB stick now
[17:17:40] <pringlescan> maybe someone did an update and didn't reboot
[17:19:09] <pringlescan> the drive isn't on rmformat or format
[17:22:20] <pringlescan> prtconf -D doesn't say anything about USB
[17:23:04] <pringlescan> oh well, I'll turn this into a training session on the datacenter… what's mounted at /usbkey then?
[17:24:45] *** alcir has quit IRC
[17:33:14] <pringlescan> well, I'm glad I found this before I left
[17:35:53] <Licenser> pringlescan I'm running dev in multiple places w/o any problme
[17:36:31] <Licenser> and /usbkey is actually on the zones pool
[17:36:52] <Licenser> it is just very stupidly named
[17:40:12] *** noahmehl has quit IRC
[18:21:03] <benoitc> Licenser: answered
[18:21:08] <benoitc> about hackney
[18:21:36] <benoitc> i could make that change anyway, will try to find some time
[19:22:22] <Licenser> benoitc :)
[19:22:47] <Licenser> it's not a pressing matter for me, I just looked thorugh fifo's pull requests and noticed that one was open so wanted to poke the bear :P
[19:26:16] <MerlinDMC> don't wake sleeping bears.
[19:54:35] <Licenser> he was just hibernating
[19:54:50] <Licenser> arekinath I think erlzdoor blocks the VM
[20:43:31] <pringlescan> so if I just put latest-usb on a stick and plug it in everything should work?
[21:23:13] *** pringlescan has quit IRC
[21:36:07] *** _lb_ has quit IRC
[21:46:55] *** pringlescan has joined #project-FiFo
[22:03:05] *** mgls has quit IRC
[22:15:48] *** pringlescan has quit IRC
[22:18:01] *** pringlescan has joined #project-FiFo
[23:05:11] *** pringlescan has quit IRC
[23:07:03] *** pringlescan has joined #project-FiFo
[23:33:53] <arekinath> Licenser: any idea where? can you send it Sig11 or attach mdb or something?
[23:35:05] <arekinath> it'd be great to know which lock...
[23:35:43] *** pringlescan has quit IRC
[23:40:15] <arekinath> although.. job open only takes one lock, the gbl->jlock
[23:40:16] <arekinath> hm
[23:47:31] <arekinath> if it's doing what I think it is, then I'm surprised it isn't segfaulting instead
[23:56:21] <arekinath> oh no, it won't get to the segfaulting stage, because it'll block first freeing the door, which then blocks out the next job_open
[23:56:22] <arekinath> right
[23:56:34] <arekinath> so this should happen every time you do an open() after a close() ?
top

   October 17, 2013  
< | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 | 28 | 29 | 30 | 31 | >