Tim Armstrong has posted comments on this change.

Change subject: IMPALA-5240: Allow configuration of number of disk I/O threads 
based on disk type
......................................................................


Patch Set 1:

(3 comments)

In addition to Matt's comments.

http://gerrit.cloudera.org:8080/#/c/7232/1/be/src/runtime/disk-io-mgr.cc
File be/src/runtime/disk-io-mgr.cc:

PS1, Line 47: num_threads_per_rotational_disk
Maybe io_threads instead of threads_ to be more specific. We can leave the 
existing num_threads_per_disk option as-is but it's probably better to match 
the below options.


PS1, Line 50: num_threads_per_flash_disk
Not your change but we probably shouldn't call this "flash", since that's a 
specific type of solid state disk. Maybe "solid_state_disk" or "ssd"?


http://gerrit.cloudera.org:8080/#/c/7232/1/be/src/runtime/disk-io-mgr.h
File be/src/runtime/disk-io-mgr.h:

PS1, Line 637:   ///  - threads_per_rotational_disk: number of read threads to 
create per rotational
             :   ///    disk. This is also the max queue depth.
             :   ///  - threads_per_flash_disk: number of read threads to 
create per flash disk. This is
             :   ///    also the max queue depth.
             :   DiskIoMgr(int num_disks, int threads_per_disk, int 
min_buffer_size,
             :       int max_buffer_size, int threads_per_rotational_disk = 0,
             :       int threads_per_flash_disk = 0)
I believe this constructor only exists for the backend tests. I think we should 
either remove the unused arguments or pass them in for at least one test. I 
think the second is better but I don't feel strongly.


-- 
To view, visit http://gerrit.cloudera.org:8080/7232
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: I094aff3747104104fe0465d73dcdbef5d9892f7c
Gerrit-PatchSet: 1
Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-Owner: Bikramjeet Vig <[email protected]>
Gerrit-Reviewer: Matthew Jacobs <[email protected]>
Gerrit-Reviewer: Tim Armstrong <[email protected]>
Gerrit-HasComments: Yes

Reply via email to