Project

General

Profile

Actions

Bug #104

closed

High io wait on Elwing

Added by Marc Dequènes almost 14 years ago. Updated over 13 years ago.

Status:
Rejected
Priority:
Urgent
Category:
System :: Hardware
Start date:
2010-06-17
Due date:
% Done:

20%

Estimated time:
Patch Available:
No
Confirmed:
No
Branch:
Entity:
DuckCorp
Security:
No
Help Needed:

Description

We are experiencing slow responses sometimes on Elwing, and it seems to affect NFS a lot.

I found high io wait lasting more than a few seconds, and sometimes several minutes long.

I looked at a bonnie++ check:

Version  1.96       ------Sequential Output------ --Sequential Input- --Random-
Concurrency   1     -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
Elwing           8G   347  95 25089   7 16206   3   397  22 58156   5  82.4   2
Latency               112ms   11751ms    4522ms    2791ms    9326ms   17265ms
Version  1.96       ------Sequential Create------ --------Random Create--------
Elwing              -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                 16  5138  10 +++++ +++  6805  11 12978  25 +++++ +++  7334  12
Latency             24172us     560us     651us    1317us     105us     252us
1.96,1.96,Elwing,1,1276727105,8G,,347,95,25089,7,16206,3,397,22,58156,5,82.4,2,16,,,,,5138,10,+++++,+++,6805,11,12978,25,+++++,+++,7334,12,112ms,11751ms,4522ms,2791ms,9326ms,17265ms,24172us,560us,651us,1317us,105us,252us

And for comparison, here is the same on Annael:

Version  1.96       ------Sequential Output------ --Sequential Input- --Random-
Concurrency   1     -Per Chr- --Block-- -Rewrite- -Per Chr- --Block-- --Seeks--
Machine        Size K/sec %CP K/sec %CP K/sec %CP K/sec %CP K/sec %CP  /sec %CP
Annael           8G   682  97 73985  13 42083   5  3548  85 19836   1 271.3   4
Latency             12457us    2543ms     349ms   67604us      347s     433ms
Version  1.96       ------Sequential Create------ --------Random Create--------
Annael              -Create-- --Read--- -Delete-- -Create-- --Read--- -Delete--
              files  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP  /sec %CP
                 16 15253  18 +++++ +++ +++++ +++ +++++ +++ +++++ +++ +++++ +++
Latency             11974us     623us     490us     966us     194us     113us
1.96,1.96,Annael,1,1276738616,8G,,682,97,73985,13,42083,5,3548,85,19836,1,271.3,4,16,,,,,15253,18,+++++,+++,+++++,+++,+++++,+++,+++++,+++,+++++,+++,12457us,2543ms,349ms,67604us,347s,433ms,11974us,623us,490us,966us,194us,113us

HD temperatures are about 43-45°C, which is a bit high, but is the same as on Daneel, which is working fine.

I need to investigate more.


Related issues 2 (0 open2 closed)

Related to DuckCorp Infrastructure - Bug #102: Elwing is BorkenResolvedMarc Dequènes2010-06-14

Actions
Related to DuckCorp Infrastructure - Enhancement #117: Plans for the futur of HQ server/NAS/...ResolvedMarc Dequènes2010-08-02

Actions
Actions #1

Updated by Marc Dequènes almost 14 years ago

  • % Done changed from 10 to 20

The situation is confirmed: the new disks are not suitable for RAID, as Western Digital have restrained their recent product in order to improve sells of their "enterprise class" products.

One of the new disks is having problems with the widle3.exe utility, and has been returned.

I'm looking for a solution...

Actions #2

Updated by Marc Dequènes over 13 years ago

I tested WDTLER too, but failed as expected with this kind of disk (EARS serie).

Actions #3

Updated by Marc Dequènes over 13 years ago

  • Status changed from In Progress to Rejected

So, it's impossible to solve the situation without changing disks. This is discussed in #117.

Actions #4

Updated by Marc Dequènes over 13 years ago

  • Category changed from System :: Base to System :: Hardware
Actions

Also available in: Atom PDF