| 03:28 | <MikeSmith> | annevk: so, about running the checker from travis, have tried a few things with my server but now coming the the conclusion that trying to run it on my server VM is not going to work |
| 03:28 | <MikeSmith> | problem is, running out of Java heap space |
| 03:29 | <MikeSmith> | it is also maxing out the CPU for extended periods of time |
| 03:29 | <MikeSmith> | so, need to move it to another server |
| 03:30 | <MikeSmith> | and once I get it set up you can update the CNAME and I think we will find it works fine |
| 03:32 | <MikeSmith> | am setting up an instance on a machine with 4 cores and 8GB of RAM (5GB of which is free most of the time) |
| 09:35 | <annevk> | MikeSmith: okay, seems that currently it's still falling over on something, but sounds promising |
| 10:31 | <MikeSmith> | annevk: yeah it was falling over because of java running out of heap space |
| 10:32 | <MikeSmith> | I had been trying to run it with only 256MB of heap but it basically seems to need at least 512MB |
| 10:33 | <MikeSmith> | so I upped it to that and for now I think that’ll be fine but the real fix is to just move it |
| 10:33 | <MikeSmith> | also I think it won’t require you to make any change to the CNAME record |
| 10:34 | <MikeSmith> | because I can keep doing the TLS termination at my server but just have nginx proxy it to the new one |
| 10:40 | <MikeSmith> | annevk: but also for the case of the Encoding spec I guess we should change the deploy script to not check the generated HTML each time but instead only if the python script has changed |
| 10:40 | <MikeSmith> | since it’s wasteful to run it when there isn ' |
| 10:41 | <MikeSmith> | *isn’t anything that would have changed the output |
| 10:44 | <MikeSmith> | plus of course if there’s a problem with the checker then it screws up your workflow |
| 10:46 | <annevk> | MikeSmith: aside: "katakana" without capital k is fine? |
| 10:46 | <annevk> | MikeSmith: yeah, if that's easy to change that sounds good |
| 10:47 | <annevk> | MikeSmith: wasting a bunch of bandwidth and CPU seems sad |
| 11:03 | <MikeSmith> | annevk: yeah "katakana" without capital k is fine |
| 12:15 | <annevk> | Ta |
| 14:10 | <MikeSmith> | annevk: OK, checker.html5.org now moved to a much beefier server |
| 14:10 | <MikeSmith> | so, no more memory problems, no more timeouts |
| 14:11 | <MikeSmith> | later will work on making the change for Encoding to only run the checker if the python script changes |
| 14:11 | <MikeSmith> | though may not get to it til next weekend |
| 15:14 | <annevk> | MikeSmith: no rush, thanks for taking the time |
| 16:16 | <Alystair> | howdy all is there a way to tell if a computer is capable of playing audio or if sound is currently muted, and if not - where would I go to suggest this for future implimentation in web spec? |
| 16:21 | <Alystair> | I feel this could be used in situations to have fall backs in case a site requires audio but the computer either has no hardware available or is actively muted outside of the scope of the frame |