10 to 20% of the time queries take the better part of a second when only on one
node and you indicate the queries are very similar in nature.
For that second do a lot of queries show up all at once or are they spaced out?
Do you see a spike in query requests?
Do you see the CPU max out?
Do you se
i would rather suggest you separate the nodes roles. you could divide them
into 3 node roles.
1.Index node : for index data
2.load balance : for query , you could maximize load balance node memory
for avoiding the GC consume too much memory.
3.data node : just for store the data.
and optimize th
Suspecting this, we tried taking things down to a single server and still
have the exact same response.
That said, optimizing the query to get rid of some of those ORs has helped,
so I think that's the path we're taking.
--
You received this message because you are subscribed to the Google Gro
ORs certainly tend to be slower than simpler/shorter term queries, but I'd
suspect that cross-DC part because your index is tinny and servers are
beefy and plentiful. Maybe you can look at your network and query metrics
and correlate a drop or spike in traffic or packet loss with slow queries?
It has been suggested that what I'm seeing is a CPU-bound issue in that the
large number of OR directives in our query could make many of these queries
take a long time.
As I'm not an expert on crafting queries, any expert opinions?
Because I'm feeling pretty good about my configuration about n
Okay, let's attack this directly. We have a cluster of 6 machines (6
nodes). We have an index of just under 3.5 million documents. Each document
represents an Internet domain name. We are performing queries against this
index to see names that exist in our index. Most queries are coming back in