I am using arangodb for creating a database for some GB of data. I have 
imported the document collections say A and B. Now I am importing edges 
between A->B(_from,_to). I am using arangoimp for this. I am importing the 
data using a shell script where it which loops over the files that needs to 
be imported. Each and every file < 11 MB. The format of the shell script is 
as follows:

for filename in /path/<folder_name>/*.csv; do
arangoimp --server.endpoint tcp://**********/ 
<http://www.google.com/url?q=http%3A%2F%2F50.17.248.223%3A8531%2F&sa=D&sntz=1&usg=AFQjCNGphQ-Y0dNupIRgceX833UidQaeIA>
 --server.username 
root --server.password ****** --file $filename --type csv --server.database 
TestDB --on-duplicate ignore --collection <collectionname> > 
"$filename"_log"
rm "$filename"_log
done

*The Problem:* Importing speed is significantly low frankly speaking its < 
1mbps. I am running it on cluster of three servers one of them I set up as 
coordinator and rest of them are dbservers. I have checked logs and it is 
showing me a warning. Here is the log for reference. 

WARNING {queries} slow query: 'FOR x IN @@collection LET att = 
APPEND(SLICE(ATTRIBUTES(x), 0, 25), '_key', true) LIMIT @offset, @count 
RETURN KEEP(x, att)', took: 14.240174

1. for document collection I have my own key(So that I can create edges 
easily with _key)
2. I have multiple edges between specific selected set of vertices, to be 
precise if v1 is a document in A and v2 is a document in B, then I may have 
5 edges between these two documents(vertices)
3. I made sure the self generated keys should not contain any character 
else of recommended ones.
4. I have self generated key in all cases collection A, B and edges.

Please help me to improve the speed or guide me what is wrong in my 
approach.

-- 
You received this message because you are subscribed to the Google Groups 
"ArangoDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to