Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • 16:39
    ejsmith synchronize #452
  • 16:39

    ejsmith on elastic7

    Log the correct task id and cat… (compare)

  • 06:25
    ejsmith synchronize #452
  • 06:25

    ejsmith on elastic7

    Minor (compare)

  • 06:00
    ejsmith synchronize #452
  • 06:00

    ejsmith on elastic7

    Add missing task id (compare)

  • Dec 13 22:43
    ejsmith synchronize #452
  • Dec 13 22:43

    ejsmith on elastic7

    More reindex job changes (compare)

  • Dec 13 15:46
    ejsmith synchronize #452
  • Dec 13 15:46

    ejsmith on elastic7

    Just retry 3 times (compare)

  • Dec 13 15:44
    ejsmith synchronize #452
  • Dec 13 15:44

    ejsmith on elastic7

    More log message changes (compare)

  • Dec 13 14:25
    niemyjski synchronize #452
  • Dec 13 14:25

    niemyjski on elastic7

    Added retry reindex for disconn… (compare)

  • Dec 13 14:09
    niemyjski synchronize #452
  • Dec 13 14:09

    niemyjski on elastic7

    Updated migration to retry sock… (compare)

  • Dec 13 04:51
  • Dec 13 03:28
    ejsmith synchronize #452
  • Dec 13 03:28

    ejsmith on elastic7

    Oops progress (compare)

  • Dec 13 03:04
    ejsmith synchronize #452
Blake Niemyjski
@niemyjski
yeah
Sander Rijken
@srijken
we want queue behavior for cleanup
for the rest we want ConcurrentDictionary behavior
Blake Niemyjski
@niemyjski
really wish I could step into these extension methods or go to source but all I see is empty definitions
yep
Sander Rijken
@srijken
_mergedEvents.FirstOrDefault() has the same issue
Blake Niemyjski
@niemyjski
yeah
Sander Rijken
@srijken
hmm what to do
Blake Niemyjski
@niemyjski
well I think the issue is that it’s not found in the queue and then other code runs then it enqueues it between checks and the lock prevents that
Sander Rijken
@srijken
the lock makes the whole thing threadsafe yeah
btw, what’s the difference between FormattedInfo and FormattedTrace?
couldn’t get the Trace to show up in the test runner
Blake Niemyjski
@niemyjski
not sure
oh
LogLevel
and there is a min level in the unit tests of Debug
could update that in GetClient
Sander Rijken
@srijken
to get rid of the lock, we would have to keep both a concurrentdictionary (for checking if we have them) and a concurrentqueue (for cleanup of the dictionary)
Blake Niemyjski
@niemyjski
or CreateClient on the plugin tests
well if we are locking do we even need concurrentqueue anymore?
Sander Rijken
@srijken
no
Blake Niemyjski
@niemyjski
and is concurrentdictionary even in the pcl?
Sander Rijken
@srijken
we could also do a reader/writer lock
Blake Niemyjski
@niemyjski
what ever you think is best and easiest to maintain :)
Sander Rijken
@srijken
:)
it’s not that much harder
Blake Niemyjski
@niemyjski
@ejsmith this stuff is pretty specific to exceptionless
hitting some roadblocks
Eric J. Smith
@ejsmith
with the aggregations?
stuff like avg:somefield min:somefield, max:somefield
that shouldn’t be specific
Blake Niemyjski
@niemyjski
i know
but we need access to the elastic index for alias name
Eric J. Smith
@ejsmith
so what is specific?
Blake Niemyjski
@niemyjski
can share my screen quick
Eric J. Smith
@ejsmith
what do you mean alias name
can’t do that.
Blake Niemyjski
@niemyjski
No generic way to inject the index on the context unless we add a constraint to the index with a <T> for the hell of it to auto inject it
then the date stuff is not generic but something we’d have to ahve
that’s only for terms tho
Screen Shot 2016-04-11 at 4.12.07 PM.png
            .WithDateRange(utcStart, utcEnd, EventIndex.Fields.PersistentEvent.Date)
            .WithIndices(utcStart, utcEnd, $"'{_eventIndex.VersionedName}-'yyyyMM”); 
is the biggest thing
you want to do smart date ranges
Sander Rijken
@srijken
@niemyjski no ReaderWriterLock in PCL it seems, so just keep the lock the way it is?
Blake Niemyjski
@niemyjski
yeah I guess
Eric J. Smith
@ejsmith
that is up to the user to pass in a filter and system filter just like any other thing in the repo
aggregations are on top of that.
Blake Niemyjski
@niemyjski
wonder if we can get away with non concurrent queues but is it worth it? what might it introduce
Eric J. Smith
@ejsmith
use the exact same query builder stuff