Threaded I/O tester / Patches / #2 fix random IO statistics reporting bug and some improvements

#2 fix random IO statistics reporting bug and some improvements

Status: open-works-for-me

Owner: Mika Kuoppala

Labels: None

Priority: 9

Updated: 2008-08-18

Created: 2008-08-12

Creator: rank

Private: No

This patch did several jobs:

1. redefine DEFAULT_RANDOM_OPS from 1000 to 1024, which made the random IO more accurate;
2. change parameters' type of update_latency_info from struct timeval to its according pointers, which runs more effeciently;
3. blocks = io_ops;

Among this the 3rd is the most important point:
1. the former code has statistics bug in reporting random IO rate (sometimes it reports an SATA has 10,000 MB/s writing rate!): the original author just misused one variable.

Discussion

rank - 2008-08-12

patch to the latest CVS head which stopped at v0.4.1 on 2008-04-18

tiobench-to-20080418.diff

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

rank - 2008-08-12

Logged In: YES
user_id=1429137
Originator: YES

it reports one SATA disk has 10,000 MB/s writing performance, do you believe it?

If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

rank - 2008-08-13

priority: 5 --> 9

assigned_to: nobody --> jmm

status: open --> open-works-for-me
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

rank - 2008-08-18

assigned_to: jmm --> miku
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:

fix random IO statistics reporting bug and some improvements

Group

Searches

Help

#2 fix random IO statistics reporting bug and some improvements

Discussion