Dgraph v1.0.10-rc1 release candidate


(Daniel Mai) #1

Hey all,

Please try out our latest release candidate v1.0.10-rc1.

Link to binaries: Linux, Darwin, Windows.

Also published on Docker tagged as v1.0.10-rc1.

Note: This is not the official release. You won’t be able to get these binaries using the get.dgraph.io quick install script.

Release notes:

  • Fixes related to membership quorum stability and memory usage.
  • Breaking change: Dgraph Server is now called Dgraph Alpha.
    • The command dgraph server has been renamed to dgraph alpha.
    • The prometheus metric dgraph_server_health_status is now called dgraph_alpha_health_status.
  • The glog library is now used for Dgraph Zero and Dgraph Alpha logging output.
    • By default, logs are written to stderr and to log files in /tmp. To only log to stderr and not to files, pass the flag --logtostderr. To specify the log output directory, use the --log_dir flag.
  • Alter operations can be protected by requiring an auth token from clients. See the documentation for more info.

Cheers,

Daniel


(Daniel Mai) #2

(Orlando) #3

Are there more changes besides the server rename that we can test?, I read in the forum that some improvements in the HAS functions were in process.


(Michel Conrado (Support Engineer)) #4

Not really, just the latest changes and what is in release notes. The “Has” function improvement still to be pushed to test soon. And There is a lot of work yet to be done. Like native Backup and so on.


#5

I got several
I1024 09:48:16.310122 1 predicate_move.go:238] Proposed 3721929 keys. Error: Server unavailable. Please retry later
during alpha instance balancing even thought that server is still running.


#6

Also after I alter predicate, I cannot find anything in this predicate.
And this simple query takes 10s

{
  e(func: has(text.value), first: 10) {
    uid
    expand(_all_)
    _predicate_
  }
}

Setup is 2x AWS EC2 m5d.xlarge


#7

Custom plug-in (tokenizer) is also not working (until now it works).

text.value: string @index(term,trigram,nfd) .

Return in ratel something like:
context deadline exceeded

And in alpha logs I can read:

I1024 10:15:12.343296       1 http.go:399] Got alter request via HTTP from 109.230.41.157:52746
I1024 10:15:12.343322       1 server.go:271] Received ALTER op: schema:"text.value: string @index(term,trigram,nfd) ." 
I1024 10:15:12.344111       1 server.go:324] Got schema: [predicate:"text.value" value_type:STRING directive:INDEX tokenizer:"term" tokenizer:"trigram" tokenizer:"nfd" ]
I1024 10:15:16.345315       1 mutation.go:164] Done schema update predicate:"text.value" value_type:STRING directive:INDEX tokenizer:"term" tokenizer:"trigram" tokenizer:"nfd" 
I1024 10:15:16.345420       1 mutation.go:180] Done schema update predicate:"text.value" value_type:STRING directive:INDEX tokenizer:"term" tokenizer:"trigram" tokenizer:"nfd" 
I1024 10:15:16.345436       1 server.go:328] ALTER op: schema:"text.value: string @index(term,trigram,nfd) ."  done

#8

I used bulk load to import data into dgraph.
I cannot find any data from bulk load, but I can find new one inserted with transaction (in text.value predicate) and also tokenizer is working.


(Daniel Mai) #9