These are chat archives for exceptionless/Discuss

29th
Mar 2016
Bill Henning
@BillHenning
Mar 29 2016 01:04 UTC
Doesn't seem like that is working right. I have a bunch of rows with 0%. What you're going to find is that most people never have exceptions (assuming your app is written well). So if 0.001% of people run into something, it's showing 0%. That's not very helpful.
Blake Niemyjski
@niemyjski
Mar 29 2016 14:33 UTC
yeah
or there isn’t a user associated to an error
(anon) and now users affected is 0
there’s still alot of work to go into it
@ejsmith the app service plan has cpu sitting at 4% and memory at 20%
I suggest we run multiple instances on the event post job
maybe bump it up to 2
?
Eric J. Smith
@ejsmith
Mar 29 2016 14:35 UTC
is it still behind?
Blake Niemyjski
@niemyjski
Mar 29 2016 14:35 UTC
no
saying we bump up the instance per dedicated event post job from 1 to 2
cpu is crazy low
memory is low
wish you could scale web jobs independently
Eric J. Smith
@ejsmith
Mar 29 2016 14:36 UTC
yeah, how was it when it was blasting through the events?
Blake Niemyjski
@niemyjski
Mar 29 2016 14:36 UTC
would be sooo nice
Eric J. Smith
@ejsmith
Mar 29 2016 14:36 UTC
yeah, I think that might be coming soon.
Blake Niemyjski
@niemyjski
Mar 29 2016 14:37 UTC
when we scaled up
Eric J. Smith
@ejsmith
Mar 29 2016 14:37 UTC
before it
Blake Niemyjski
@niemyjski
Mar 29 2016 14:37 UTC
Screen Shot 2016-03-29 at 9.36.51 AM.png
the hover is on 8am when I scaled up
you can see my deploy this week really made the cpu go down :)
Eric J. Smith
@ejsmith
Mar 29 2016 14:38 UTC
ok, so we aren’t maxing out what we have.
Blake Niemyjski
@niemyjski
Mar 29 2016 14:38 UTC
not even close but we need the core count (4) that comes with our app service plan
or we could scale that down
Eric J. Smith
@ejsmith
Mar 29 2016 14:38 UTC
so yeah, lets try changing the instance count to 2
Blake Niemyjski
@niemyjski
Mar 29 2016 14:38 UTC
ok
Eric J. Smith
@ejsmith
Mar 29 2016 14:39 UTC
yeah, I really want jobs to be managed separately.
it seems like such a no brainer to me that you should be able to run scripts, exes and js as a 1st class thing.
but even AWS doesn’t have that.
but they really don’t do any PaaS type stuff
Blake Niemyjski
@niemyjski
Mar 29 2016 14:40 UTC
:(
could be a killer feature for azure
Eric J. Smith
@ejsmith
Mar 29 2016 14:41 UTC
it looks like heroku has it.
and you can scale them separately.
Blake Niemyjski
@niemyjski
Mar 29 2016 14:42 UTC
just need to hammer scott gu :)
could really make a lot of people happy
Eric J. Smith
@ejsmith
Mar 29 2016 14:42 UTC
I think its coming.
Blake Niemyjski
@niemyjski
Mar 29 2016 14:43 UTC
omg so that drop in cpu from that picture I sent you
Eric J. Smith
@ejsmith
Mar 29 2016 14:43 UTC
I saw some hints.
Blake Niemyjski
@niemyjski
Mar 29 2016 14:43 UTC
is just from those stats changes and caching project count
Eric J. Smith
@ejsmith
Mar 29 2016 14:43 UTC
nice
man, heroku is so damn nice.
.net core on heroku will be sweet.
this .net core crap needs to hurry up.
Blake Niemyjski
@niemyjski
Mar 29 2016 14:51 UTC
yeah
hehe
Eric J. Smith
@ejsmith
Mar 29 2016 14:51 UTC
so many awesome hosting options that we can’t have.
.net is like 5 years behind the other platforms.
at least
problem is that I really like the language. The platform sucks, but the language is really nice.
.net core will be a massive step towards correcting that.
Blake Niemyjski
@niemyjski
Mar 29 2016 15:12 UTC
    public async Task<JobResult> RunAsync(CancellationToken cancellationToken = default(CancellationToken)) {
        var linkedCancellationTokenSource = CancellationTokenSource.CreateLinkedTokenSource(cancellationToken, TimeSpan.FromSeconds(30).ToCancellationToken());
dang
so looking at this workitem job
doesn’t that mean we can never renew it and it will auto cancel after 30 seconds?
Eric J. Smith
@ejsmith
Mar 29 2016 15:13 UTC
which one?
Blake Niemyjski
@niemyjski
Mar 29 2016 15:13 UTC
WorkItemJob
was looking into why we were getting cancellation message
Eric J. Smith
@ejsmith
Mar 29 2016 15:13 UTC
well that isn’t for event processing though.
Blake Niemyjski
@niemyjski
Mar 29 2016 15:13 UTC
I know
but I saw an error and we had 30 of them
so I looked into why it would be failing
it’s not failing it’s being cancelled
wish it was more clear on what type was cancelling it
Eric J. Smith
@ejsmith
Mar 29 2016 15:14 UTC
yeah, need to know what type of work item it is.
need to improve the logging around that.
Blake Niemyjski
@niemyjski
Mar 29 2016 15:16 UTC
yeah
Blake Niemyjski
@niemyjski
Mar 29 2016 15:25 UTC
so if it takes longer than 30 seconds to dequeue a work item we cancel
that doesn’t seem right at all
that dequeue is efficient
we should only bail if cancellation was requested by the original token
only the dequeue method should be using the linked token
I’ve updated the code,
pushed
nice find :D
Blake Niemyjski
@niemyjski
Mar 29 2016 15:45 UTC
@ejsmith when you think we are solid and done breaking stuff we should do a foundatio 4 release
:D
Eric J. Smith
@ejsmith
Mar 29 2016 15:45 UTC
what did you change?
yeah, I think those changes look good.
Blake Niemyjski
@niemyjski
Mar 29 2016 16:18 UTC
yeah
:)
guess all our work item jobs are short lived so we never ran into it
I just added a warning message anytime an event is cancelled in the pipeline
want to see why these are getting cancelled
I dreamed about aurelia last night
I really wonder what it would take to port our app over to it… maybe some day we would know.. considering we are following conventions for a lot of stuff and we have controllers with plan javascript I don’t think it would take too much if you wrote custom code to use those conventions and used the angular databinding syntax
Frank Ebersoll
@frankebersoll
Mar 29 2016 16:23 UTC
hey there!
Blake Niemyjski
@niemyjski
Mar 29 2016 16:23 UTC
pretty cool to watch where they are taking it and all the updates
hey
Frank Ebersoll
@frankebersoll
Mar 29 2016 16:23 UTC
@niemyjski haven’t seen this ZeroDivision exception yet
is it reproducable?
Blake Niemyjski
@niemyjski
Mar 29 2016 16:24 UTC
link?
I’m not sure. it should be repeatable I hope
Eric J. Smith
@ejsmith
Mar 29 2016 16:32 UTC
hey @frankebersoll
:D
Worked my magic
Frank Ebersoll
@frankebersoll
Mar 29 2016 17:32 UTC
found this http://www.dylanbeattie.net/2016/03/how-to-really-break-internet.html on morning brew some days ago… you probably already talked about that :-)
Blake Niemyjski
@niemyjski
Mar 29 2016 17:34 UTC
@ejsmith how come the throttled lock provider doens’t take a messagebus but the cachelock provider does
yeah
:D
Can’t reproduce over here :-(
maybe there’s invalid keys in your storage, but it should just gracefully delete them.
Blake Niemyjski
@niemyjski
Mar 29 2016 17:52 UTC
yeah it was something weird
the server was down
and it was from a fresh page load
kinda crazy
i tried a few times but didn’t see it
@ejsmith I updated the foundatio readme
can you take a look and update it any
and lets get a 4.x release out there now that it’s on the ms blog
think we should add logging there
I’ll work on a release notes
but let me know
Eric J. Smith
@ejsmith
Mar 29 2016 19:41 UTC
just getting the readme completely up to date is fine.
don’t think we need to do another major release.
what we need to do is get our process down for beta builds and then major official releases will make a lot more sense.
Blake Niemyjski
@niemyjski
Mar 29 2016 20:50 UTC
yeah
Thanks @jamie94bc and everyone else who helped :) https://github.com/exceptionless/Foundatio/releases/tag/v4.0.0
Jamie Clarke
@jamie94bc
Mar 29 2016 20:53 UTC
No problem! Wish I had the time to help more!
looks like they backported it
it’s all good :)
new insiders build out