These are chat archives for exceptionless/Discuss

12th
Feb 2016
Blake Niemyjski
@niemyjski
Feb 12 2016 00:07 UTC
@ejsmith why didn’t you use the existing metric tests for redis?
and you forgot to update the readme :)
Eric J. Smith
@ejsmith
Feb 12 2016 00:08 UTC
There really wasn’t any.
and it’s hard to test metrics in a standard way
because the contracts have no way to retrieve values
nor should they
hrmm… I just updated the repos project to use the latest 4.0 Foundatio, but it still lists requirement as Foundatio 3.0 <4.0
Blake Niemyjski
@niemyjski
Feb 12 2016 00:12 UTC
hmm
hard coded?
Eric J. Smith
@ejsmith
Feb 12 2016 00:13 UTC
don’t see it hard coded anywhere.
Blake Niemyjski
@niemyjski
Feb 12 2016 00:13 UTC
I’m looking
fixed
Eric J. Smith
@ejsmith
Feb 12 2016 00:15 UTC
where?
I was looking at the wrong file
Blake Niemyjski
@niemyjski
Feb 12 2016 00:15 UTC
lol
kinda excited
Eric J. Smith
@ejsmith
Feb 12 2016 00:17 UTC
about?
Blake Niemyjski
@niemyjski
Feb 12 2016 00:17 UTC
guys released new version of htop
version 2.0!
Eric J. Smith
@ejsmith
Feb 12 2016 00:17 UTC
lol
Blake Niemyjski
@niemyjski
Feb 12 2016 00:17 UTC
and it added official support for osx
not on brew yet
but added mouse support and a bunch of stuff
Eric J. Smith
@ejsmith
Feb 12 2016 00:18 UTC
yeah, why isn’t it on brew damn it.
Blake Niemyjski
@niemyjski
Feb 12 2016 00:18 UTC
:(
I love brew
I have a command I run once a day
brew update && brew upgrade && ncu -g
updates stuff and lets me know if my npm global packages are out of date
:D
bleeding edge
Eric J. Smith
@ejsmith
Feb 12 2016 00:19 UTC
wow
Blake Niemyjski
@niemyjski
Feb 12 2016 00:19 UTC
got an addiction to be bleeding edge
@ejsmith we’ll review what I’m working on and look at the pr for manual stacking
I think it’s good to go
Eric J. Smith
@ejsmith
Feb 12 2016 00:23 UTC
boom htop 2.0
lol
Sander Rijken
@srijken
Feb 12 2016 00:32 UTC
@niemyjski still need to do the extra testing for deduping
Blake Niemyjski
@niemyjski
Feb 12 2016 00:35 UTC
Ok
Would be kind of cool if we had an experiment client package
Sander Rijken
@srijken
Feb 12 2016 00:36 UTC
yeah
Blake Niemyjski
@niemyjski
Feb 12 2016 00:36 UTC
Where users could test stuff and give feedback and then things gradually merge into the real client
Sander Rijken
@srijken
Feb 12 2016 00:36 UTC
well I guess you can push a prerelease to nuget
Blake Niemyjski
@niemyjski
Feb 12 2016 00:36 UTC
Or go away
Yeah
Kinda like vs web packages where they work on new things then merge them into core vs
Btw :)
Blake Niemyjski
@niemyjski
Feb 12 2016 03:26 UTC
hmm
welp thats coming along :) just need to do some edge cases and parameter parsing.. got unit tests now
I know I do some stupid stuff from time to time, but I’m just dumbfounded on the shier amount of stupid crap I see in the raygun spec alone.. HUGE naming inconsistencies that lead to extra serialization work, incomplete models which cause you to parse parameter info on a per language basis, all data is non normalized (each platform specifies the memory size could be in bytes, KB or MB)
like little things you do when writing a platform that makes everything easier on you..
and they didn’t do a single dam one of them
some things I found that I want… I want to capture locale, browser dimensions, and color depths
but we are going to get that with sessions v2..
sheer*
Jamie Clarke
@jamie94bc
Feb 12 2016 09:13 UTC
how are you planning to display that info from the redis metric addition @ejsmith ?
was just starting to look at using librato and you're moving away from it!
the pricing page makes it look pretty cheap?
their webhooksstore returns an enum to indicate failure :cry:
Eric J. Smith
@ejsmith
Feb 12 2016 13:10 UTC
@jamie94bc I am going to build an API endpoint that exposes the metrics. Then we can build a UI on top of that.
Jamie Clarke
@jamie94bc
Feb 12 2016 13:12 UTC
cool!
i need to implement something to monitor API version usage! :sweat:
Eric J. Smith
@ejsmith
Feb 12 2016 13:17 UTC
I just wanted something simpler and that can be included in the Exceptionless UI.
Library does some weird things with their stats too.
Librato
Blake Niemyjski
@niemyjski
Feb 12 2016 15:13 UTC
@jamie94bc could do that pretty easy with an owin pipeline handler
Jamie Clarke
@jamie94bc
Feb 12 2016 15:14 UTC
yeah - pretty simple it's just where to log it too (with pretty dashboards and without writing something myself) :smile:
Blake Niemyjski
@niemyjski
Feb 12 2016 15:14 UTC
well if you’d be willing to help test our deduping plugin you might have a big win with exceptionless
we are not currently rolling up metrics (combining) but we are almost there
so you’d use a lot less events and get the same value
but I need to finish custom dashboards (multiple different phases and I’m working on step one), that would allow you to view those
well I guess you could look at at feature usages but not the sum of counts on a timeline right now
Blake Niemyjski
@niemyjski
Feb 12 2016 15:26 UTC
htop 2.0 is on brew dang that was quick
Eric J. Smith
@ejsmith
Feb 12 2016 15:36 UTC
@jamie94bc yeah, eventually I want to just have an Exceptionless metrics client and be able to report all my metrics there.
Jamie Clarke
@jamie94bc
Feb 12 2016 17:28 UTC
might have to do something about this!
blob
need a warning but it's called from the queue job
i'm fixing it for now by using AzureStorageQueue instead :laughing:
something i've been meaning to do anyway
Eric J. Smith
@ejsmith
Feb 12 2016 17:33 UTC
cool
Jamie Clarke
@jamie94bc
Feb 12 2016 17:34 UTC
i think ideally the queue job should use startworking?
at the moment it's polling manually
some duplicate functionality there - but using startworking wouldn't allow us to return a job result
that was so nice and easy to change :+1:
has anyone done this in a production environment before? I assume you would create a new job (or upgrade the current and create a legacy job) to process all the stuff remaining in the old implementation
Blake Niemyjski
@niemyjski
Feb 12 2016 17:38 UTC
yeah
we haven’t really done that, we just have new jobs but I guess you’d create two jobs
push to migrate the old stuff then just delete it
Eric J. Smith
@ejsmith
Feb 12 2016 17:39 UTC
so the queue processor job does a dequeue which will efficiently wait.
the job itself runs in a loop
Jamie Clarke
@jamie94bc
Feb 12 2016 17:40 UTC
yeah - not sure if OnMessage in service bus is any more efficient than a long poll
but since Azure SB doesn't like cancellation tokens seq is full of those :-1:
Eric J. Smith
@ejsmith
Feb 12 2016 17:41 UTC
hmm
Jamie Clarke
@jamie94bc
Feb 12 2016 17:42 UTC
really the azure team should add a timespan and cancellationtoken overload
Eric J. Smith
@ejsmith
Feb 12 2016 17:42 UTC
well, if they just had cancellationtoken then they could support both since cancellationtokens can be timed.
Jamie Clarke
@jamie94bc
Feb 12 2016 17:42 UTC
timespan to control the wait, cancellation token to terminate the wait before the timeout expires
based on the response from the issue we raised on the storage sdk a speedy response for service bus doesn't seem likely
Eric J. Smith
@ejsmith
Feb 12 2016 17:43 UTC
you can do a linked token to do pretty much whatever you want.
Jamie Clarke
@jamie94bc
Feb 12 2016 17:43 UTC
unless maybe you have paid support
Eric J. Smith
@ejsmith
Feb 12 2016 17:43 UTC
hehe… yeah, not likely.
we can get @niemyjski on them.
he is really good at trolling them until they relent.
lol
Blake Niemyjski
@niemyjski
Feb 12 2016 17:46 UTC
link me
I’m pretty good at getting what I want
:)
just gotta email the right people
Eric J. Smith
@ejsmith
Feb 12 2016 17:46 UTC
release the troll!
Blake Niemyjski
@niemyjski
Feb 12 2016 17:46 UTC
I should create a service where you pay a small fee and then An army all +1 or comments on an issue for a small cut
lol
microsofts like I just got 10k emails from legit people asking for this
or maybe if you want your cause supported you have to pay entry by supporting another valid cause
Jamie Clarke
@jamie94bc
Feb 12 2016 17:50 UTC
this one!Azure/azure-storage-net#228
:laughing:
Jamie Clarke
@jamie94bc
Feb 12 2016 17:56 UTC
wow
you really should charge for that @niemyjski
Blake Niemyjski
@niemyjski
Feb 12 2016 17:56 UTC
if all else fails
:)
bkeepers/dear-github@4afa490
@jamie94bc they responded
lol
unreal
if the product team checked github more than once a week or not at all
we wouldn’t need to be contacting you now would we

rocketscience

Blake Niemyjski
@niemyjski
Feb 12 2016 18:25 UTC
Screen Shot 2016-02-12 at 12.24.46 PM.png
@ejsmith
Screen Shot 2016-02-12 at 12.25.05 PM.png
the stack with error:webapigeterror is the one with the manual stacking key
I think it’s be cause we didn’t roll up any stacking info because you supplied your own key
Blake Niemyjski
@niemyjski
Feb 12 2016 18:32 UTC
@adamzolotarev manual stacking is deployed :)
Now we need to get the client releases out hopefully next week.
Could proabably push .net client out later today without deduping
yeah some good changes maybe I’ll try and get that out today
Eric J. Smith
@ejsmith
Feb 12 2016 19:02 UTC
yeah, we probably want to try and do something better with the stack title.
Wondering if we should just should the Stacking Key in the Title?
that could get weird.
Blake Niemyjski
@niemyjski
Feb 12 2016 19:04 UTC
yeah
especially if they send us a hash
guess we could set the target info stuff on the collection to get you that nice info
but not use it for stacking ?
I moved that code up because it was a lot less work