Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Activity
  • 18:11
    Aaronontheweb commented #4742
  • 17:03
    Horusiath commented #4742
  • 16:10

    Aaronontheweb on dev

    Update cluster-overview.md (#47… (compare)

  • 16:10
    Aaronontheweb closed #4743
  • 16:10
    Aaronontheweb labeled #4743
  • 15:55
    to11mtm commented #4742
  • 14:16
    brah-mcdude opened #4743
  • 06:11
    Horusiath commented #4742
  • 06:07
    Horusiath commented #4742
  • Jan 23 19:44
    to11mtm synchronize #4742
  • Jan 23 19:22
    to11mtm commented #4742
  • Jan 23 18:29
    to11mtm commented #4737
  • Jan 23 18:16
    to11mtm commented #4736
  • Jan 23 18:11
    to11mtm edited #4736
  • Jan 23 18:00
    to11mtm review_requested #4742
  • Jan 23 17:59
    to11mtm edited #4742
  • Jan 23 17:58
    to11mtm review_requested #4742
  • Jan 23 17:52
    to11mtm opened #4742
  • Jan 23 03:10
    seungyongshim edited #4733
  • Jan 22 18:02
    IgorFedchenko closed #4738
Aaron Stannard
@Aaronontheweb
in terms of the Postgres HOCON
what are the table names being used?
Bart de Boer
@boekabart
default
snapshot_store and event_journal
Aaron Stannard
@Aaronontheweb
I wonder if the issue is the metadata table
but that's very odd
and also, a reason why we need akkadotnet/akka.net#3811 back
this crap would immediately power vomit an exception since the snapshot store couldn't deserialize those events
rather than YOLO serialization, which allows this type of error to silently fail in the background without any clear explanation
/ rant off
Bart de Boer
@boekabart
metadata table is empty btw
Aaron Stannard
@Aaronontheweb
can't be that then
first thing I'd look at is how the snapshot store is reading from the rows
because that's truly weird
can you file an issue on that project?
Bart de Boer
@boekabart
postgres project?
Aaron Stannard
@Aaronontheweb
correct
Bart de Boer
@boekabart
check
Aaron Stannard
@Aaronontheweb
since it seems to be specific to the Postgres driver
Bart de Boer
@boekabart
I see nothing weird in the postgres code though
Aaron Stannard
@Aaronontheweb
could be an issue with Akka.Persistence.Sql.Common
so maybe log an issue on the main Akka.NET project isntead
Bart de Boer
@boekabart
Except the postgres is on 1.3.9 but that shouldn't be an issue
Aaron Stannard
@Aaronontheweb
that will get more visibility at least
Bart de Boer
@boekabart
Will do
I tried to add some Console.Writeline to the postgres code but after building, I couldn't get that dll to work with the rest
dotnet version issues I guess
Aaron Stannard
@Aaronontheweb
well, we did make some important fixes to Akka.Persistence.Sql common in 1.3.12, 1.3.13, and 1.3.14
none for what you're describing really
serialization fixes and bug fixes for the CircuitBreaker
Bart de Boer
@boekabart
The app is using 1.3.14
Aaron Stannard
@Aaronontheweb
the plugin might need to be upgraded
Bart de Boer
@boekabart
ah, but maybe not common 1.3.14, let me check
Aaron Stannard
@Aaronontheweb
it definitely needs to be upgraded for the serialization fixes
which affected things like Cluster.Sharding
Bart de Boer
@boekabart
Gotta love good logging:
| Assembly (N=189)                                           | Version     |
|------------------------------------------------------------+-------------|
| Akka                                                       | 1.3.14.0    |
| Akka.Logger.log4net                                        | 1.3.1.0     |
| Akka.Persistence                                           | 1.3.14.0    |
| Akka.Persistence.PostgreSql                                | 1.3.9.0     |
| Akka.Persistence.Sql.Common                                | 1.3.14.0    |
| Akka.Persistence.Sqlite                                    | 1.3.14.0    |
Aaron Stannard
@Aaronontheweb
nice
been meaning to add something like that to Petabridge.Cmd
Bart de Boer
@boekabart

But I don't think that means a rebuild/fix to PostgreSql would be needed right?

Btw we switched to Sqlite for a while, which works 100% ok

Aaron Stannard
@Aaronontheweb
yeah, that looks exactly right
Bart de Boer
@boekabart
I'm writing a fresh one in that project now, maybe if indeed it turns out related, they can be linked.
Aaron Stannard
@Aaronontheweb
sounds good
but what you're describing sounds like that type of side effect
Bart de Boer
@boekabart
Aaron Stannard
@Aaronontheweb
ty
Bart de Boer
@boekabart
... and 2 issues on akka.net wrt Config, while I'm at it.
Bart de Boer
@boekabart
@Aaronontheweb @Horusiath I found a clear correlation between smth in the logging, and the recovery leading to a phantom SnapshotOffer.
In all the BAD cases, these two lines are immediately followed by the Snapshot offer
akka.persistence.journal.postgresql] Started (Akka.Persistence.PostgreSql.Journal.PostgreSqlJournal)
akka.persistence.snapshot-store.postgresql] Started (Akka.Persistence.PostgreSql.Snapshot.PostgreSqlSnapshotStore)
and only after THAT, the ReplayFilter creation is logged:
[akka://TitleCatalogizerPPTEST/system/akka.persistence.journal.postgresql/$a] Started (Akka.Persistence.Journal.ReplayFilter)
[akka://TitleCatalogizerPPTEST/system/akka.persistence.journal.postgresql] now supervising akka://TitleCatalogizerPPTEST/system/akka.persistence.journal.postgresql/$a
Aaron Stannard
@Aaronontheweb
@boekabart we might transfer those config issues to the stand-alone HOCON repo