[jira] [Commented] (COUCHDB-2828) couch_stream doesn't set an io_priority

["ASF GitHub Bot (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/COUCHDB-2828?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14934317#comment-14934317 ] 

ASF GitHub Bot commented on COUCHDB-2828:
-----------------------------------------

GitHub user tsutsu opened a pull request:

    https://github.com/apache/couchdb-couch/pull/105

    Inherit io_priority from the process creating the stream

    `couch_stream` should be able to reuse the `io_priority` of whatever process creates it; it's then the creator's responsibility to decide what `io_priority` the stream should have.
    
    Assigning streaming IO operations their own separately-tunable category of `io_priority` is an alternative to this change. The usefulness of that would depend on whether streaming IO in general has predictable load impacts that would be useful to observe and control for. Since `couch_stream` is only currently used by `couch_att`, though, giving attachment upload operations specifically an `io_priority` category might make more sense.
    
    COUCHDB-2828
    BugzID: 25815

You can merge this pull request into a Git repository by running:

    $ git pull https://github.com/cloudant/couchdb-couch 25815-couch-stream-io-priority

Alternatively you can review and apply these changes as the patch at:

    https://github.com/apache/couchdb-couch/pull/105.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

    This closes #105
    
----
commit fc3a8e2f3ef1784cb592aedf9733c7d8b7c2106d
Author: Levi McAuley <le...@leviaul.com>
Date:   2015-09-28T22:20:18Z

    Inherit io_priority from the process creating the stream
    
    COUCHDB-2828
    BugzID: 25815

----


> couch_stream doesn't set an io_priority
> ---------------------------------------
>
>                 Key: COUCHDB-2828
>                 URL: https://issues.apache.org/jira/browse/COUCHDB-2828
>             Project: CouchDB
>          Issue Type: Bug
>      Security Level: public(Regular issues) 
>          Components: Database Core
>            Reporter: Levi McAuley
>            Priority: Minor
>
> The couch_stream gen_server makes several calls to couch_file:append_binary/2 in its callbacks, but doesn't ever set io_priority in its process dictionary. This seems to result in all IO from couch_stream being handled through the "other" channel, where it's not very easily manipulable.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)