node.js client search timeout doesn't work (for me)

[liorg2]

i have a heavy query, with complicated script, and i want to give this query larger timeout.

but no matter, what number i put there, the timeout happens after 30 seconds.

here is the code:

var queryBegin = moment();
client.search({
index: strIndices,
type: "xxx",
ignoreUnavailable: true,
timeout: 600000,
body: body

}, function (error, response) {
var queryEnd = moment();
console.log("query callback. duration: " +(queryEnd-queryBegin) + " ms ");

and the console shows:
query callback. duration: 30009 ms
Request Timeout after 30000ms

thanks a lot
Lior

[spalger]

Sorry for the confusion, but the timeout parameter represents the parameter from elasticsearch's search API. It does not impact the timeout that the client imposes on requests.

Use the requestTimeout: parameter to change the timeout imposed by the client.

[liorg2]

thanks a lot for quick reply.

so i should use them both ?

timeout: 60000,
requestTimeout: 60000,

On Fri, Feb 13, 2015 at 7:17 AM, Spencer notifications@github.com wrote:

Sorry for the confusion, but the timeout parameter represents the
parameter from elasticsearch's search API. It does not impact the timeout
that the client imposes on requests.

Use the requestTimeout: parameter to change the timeout imposed by the
client.

—
Reply to this email directly or view it on GitHub
#186 (comment)
.

[spalger]

I don't see any issue in setting both :)

[raghuugare]

What is the meaning of requestTimeout: Infinity ??
We have a task of parsing a huge CSV & populating elastic search (nearly a 100 million records) & were facing a similar issue...

[charlie-s]

@raghuugare how did you resolve?

[raghuugare]

We set the requestTimeout param to a large value like 300000 & also used batch-wise processing of records.

[amir-rahnama]

@liorg2 @spalger what's the point of setting both? Doesn't the requestTimeout always override the other?

[spalger]

@ambodi, yes, but you may want to set a shorter timeout for elasticsearch so that it can still respond even if a few shards a re taking too long. Because of the way that elasticsearch is distributed, this timeout gets sent to each shard and is actually applied locally. If all of the shards respond quickly, but one is acting up, the request to that shard could timeout and elasticsearch could respond with partial results. If instead you relied on client-side timeouts then when the timeout is reached the request to elasticsearch would just be aborted and you would receive no results. Make sense?

[delvedor]

Hi! I'm closing this due to inactivity, if it is still an issue, please feel free to reopen! :)