These are chat archives for exceptionless/Discuss

29th
Sep 2015
Blake Niemyjski
@niemyjski
Sep 29 2015 15:56
man google has some great javascript
google docs keeps freezing my safari
Blake Niemyjski
@niemyjski
Sep 29 2015 16:39
woot
restoring whole cluster backups to our dev nodes
kinda excited to profile it against the whole data set :D
what will we find
Damiaan
@dampee
Sep 29 2015 16:41
how many GBytes are that?
Blake Niemyjski
@niemyjski
Sep 29 2015 16:41
each node has 1tb drive
a few hundred gbs compressed
not sure of total size as never really looked but it’s over 300gb for sure
were growing by atleast 800k events month after month
Damiaan
@dampee
Sep 29 2015 16:43
nice. That means that your dev cluster will now be costing a whole lot more :-)
Blake Niemyjski
@niemyjski
Sep 29 2015 16:43
na it’s a vm
lol
set price
Damiaan
@dampee
Sep 29 2015 16:44
luckely. A price increase would have been inevitable otherwise
Blake Niemyjski
@niemyjski
Sep 29 2015 16:44
our current live cluster is using <10% cpu and about 55% jvm memoery and 8iops
sitting there…. should be so dam fast
I have a little command center here right now :)
Damiaan
@dampee
Sep 29 2015 16:46
All logs running over screen you mean?
(yelling) screenshot, screenshot, screenshot, ... :-)
Blake Niemyjski
@niemyjski
Sep 29 2015 16:48
yeah
lol
gitter doesn’t have screenshots
Damiaan
@dampee
Sep 29 2015 16:49
twitter has :-p
Blake Niemyjski
@niemyjski
Sep 29 2015 16:49
lol
yeah
can’t wait for this restore to be completed and profile this
Damiaan
@dampee
Sep 29 2015 16:50
what are you going to profile?
Blake Niemyjski
@niemyjski
Sep 29 2015 16:50
we made massive changes for perf/async and noticed a query was taking really long last night and it shouldn’t have
Damiaan
@dampee
Sep 29 2015 16:51
ah that way
Blake Niemyjski
@niemyjski
Sep 29 2015 16:51
like 5+ seconds for most recent, which is just newest exceptions ordered by date..
Damiaan
@dampee
Sep 29 2015 16:51
That's a lot
Blake Niemyjski
@niemyjski
Sep 29 2015 16:51
but the one that does aggregations was instant lol
yeah
we want to speed everything up
across the whole site
Damiaan
@dampee
Sep 29 2015 16:52
looking forward to it!
Blake Niemyjski
@niemyjski
Sep 29 2015 16:54
me too
always something :sa:
Damiaan
@dampee
Sep 29 2015 16:55
So, basically, you are restoring the live cluster to dev because you don't know what was going on :-D
You hope to reproduce the issue
Blake Niemyjski
@niemyjski
Sep 29 2015 16:55
yeah
Damiaan
@dampee
Sep 29 2015 16:55
and if you don't find the issue you are in deep troubles
Blake Niemyjski
@niemyjski
Sep 29 2015 16:55
with large amounts of data
Damiaan
@dampee
Sep 29 2015 16:56
but because you have lots of data on the dev cluster, playing around with the DEV environment, playing will be more joyfull :-)
Blake Niemyjski
@niemyjski
Sep 29 2015 16:56
I think that we’ll see it with tons of random data
our sets in dev is what ever we generate so small amounts of unique data
yeah
Damiaan
@dampee
Sep 29 2015 16:57
We do that from time to time on our product too.
it's just so much easier to pinpoint real problems
Blake Niemyjski
@niemyjski
Sep 29 2015 16:57
yeah :)
I need to setup our dev environment to automatically restore backups once a week
do you use elastic search
Damiaan
@dampee
Sep 29 2015 16:58
no
it's all SQL for me
If we grow a bit, we might look into ES or solr (or ...)
but currently SQL is fast enough
it's only 600 MB max database
Blake Niemyjski
@niemyjski
Sep 29 2015 17:00
yeah
sql azure is slow as heck
Damiaan
@dampee
Sep 29 2015 17:00
I know
Blake Niemyjski
@niemyjski
Sep 29 2015 17:00
we have a small prod website using it and omg is it slow
Damiaan
@dampee
Sep 29 2015 17:00
I cache a lot :-)
Blake Niemyjski
@niemyjski
Sep 29 2015 17:00
cache all the things
you using foundatio :D?
Damiaan
@dampee
Sep 29 2015 17:01
No. I looked into it once
But was wondering how I could integrate it into an application that exists for 3+ years
lots of "foundation" has already been written
Blake Niemyjski
@niemyjski
Sep 29 2015 17:02
well you don’t have to use everything just pick the peices
Damiaan
@dampee
Sep 29 2015 17:02
true
Blake Niemyjski
@niemyjski
Sep 29 2015 17:02
first step would be to change a part of it like caching
what do you use for caching? redis?
Damiaan
@dampee
Sep 29 2015 17:03
Only in memory currently
Blake Niemyjski
@niemyjski
Sep 29 2015 17:04
we have some good unit tests for our stuff
we found issues early on with in memory cache where if you called set or get with a reference type and then updated the value the cache would be updated.
so we do a copy using expression trees
Damiaan
@dampee
Sep 29 2015 17:05
neat
Blake Niemyjski
@niemyjski
Sep 29 2015 17:06
wonder if your running into anything like that now :worried:
Damiaan
@dampee
Sep 29 2015 17:06
No problems with that
never had in fact
Blake Niemyjski
@niemyjski
Sep 29 2015 17:06
that’s good :)
okay time for lunch
Damiaan
@dampee
Sep 29 2015 17:07
I was very interested in the statsd things
have a good meal
cya
Blake Niemyjski
@niemyjski
Sep 29 2015 17:07
yeah, we use that :)
Damiaan
@dampee
Sep 29 2015 17:07
where is it hosted?
Blake Niemyjski
@niemyjski
Sep 29 2015 17:08
librato
if you haven’t implemented it yet check it out we use foundatio for the abstraction
the in memory stats client will output throughput information to the output window and then you can just flip a di injection and use statsd
Damiaan
@dampee
Sep 29 2015 17:09
I will, That was the reason why I was looking at foundatio
Blake Niemyjski
@niemyjski
Sep 29 2015 17:09
we want to get foundatio using vnext very soon too
ok brb
Damiaan
@dampee
Sep 29 2015 17:09
see you