Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

i have a server with high load average, I think the problem is the raid 1.

Server is in one or two points of load average, and in seconds jumps to 10 points (md4_raid1 process), this happens once or twice per minute

cat /proc/mdstat  Personalities : [raid1]  md0 : active raid1 sdb1[1] sda1[0]
      256896 blocks [2/2] [UU]
       md3 : active raid1 sdb3[1] sda3[0]
      2562240 blocks [2/2] [UU]
       md4 : active raid1 sdb5[1] sda5[0]
      958566272 blocks [2/2] [UU]
       md1 : active raid1 sdb2[1] sda2[0]
      15366080 blocks [2/2] [UU]

model name : Intel(R) Core(TM)2 Duo CPU E8400 @ 3.00GHz

Linux 2.6.18-164.6.1.el5.centos.plus (local)   04/19/2010

avg-cpu:  %user   %nice %system %iowait  %steal   %idle
          17.37    0.01    6.02   26.17    0.00   50.43

Device:            tps   Blk_read/s   Blk_wrtn/s   Blk_read   Blk_wrtn
sda              61.09       562.65       893.73    1557214    2473546
sda1              0.01         0.27         0.02        751         42
sda2              6.11       195.50       169.78     541075     469888
sda3              0.01         0.23         0.00        641          0
sda4              0.00         0.01         0.00         18          0
sda5             54.96       366.54       723.94    1014449    2003616
sdb              54.40       433.22       893.73    1199015    2473546
sdb1              0.01         0.16         0.02        436         42
sdb2              5.31       169.00       169.78     467729     469888
sdb3              0.01         0.31         0.00        865          0
sdb4              0.00         0.00         0.00         10          0
sdb5             49.05       263.65       723.94     729695    2003616
md1              29.96       364.39       166.68    1008498     461312
md4             124.15       630.07       713.28    1743822    1974112
md3               0.05         0.43         0.00       1192          0
md0               0.04         0.32         0.00        872         10
dm-0              7.96        83.29        23.02     230530      63720
dm-1              3.67        51.81         2.73     143394       7560
dm-2              7.63        67.76        27.35     187546      75696
dm-3              8.20       134.60        14.02     372514      38792
dm-4              5.90        10.66        39.35      29498     108912
dm-5             17.39        24.52       121.79      67850     337080
dm-6             27.19       229.60       139.89     635442     387168
dm-7              0.14         1.07         0.28       2970        776
dm-8             25.84         4.23       202.89      11698     561536
dm-9             14.77         8.38       112.35      23202     310960
dm-10             5.29        12.78        29.55      35376      81784
dm-11             0.16         1.25         0.05       3450        128

the server runs lvm in the md4 cPanel runing apache, mysql, exim, dovecot, pure-ftpd with 600 accounts with WD 1TB 7200 Disks.

share|improve this question
add comment

2 Answers

up vote -1 down vote accepted

you should use hardware raid

share|improve this answer
    
This isn't a particularly useful answer. Putting his two WD 1TB drives onto a hardware RAID controller isn't likely to improve performance substantively, if at all. He needs a much better disk IO system to handle this load. –  Daniel Lawson Aug 19 '10 at 1:28
add comment

You didn't really provide enough information to thoroughly answer your question. For example:

  • What does the server do? Web server? Database server? File Server? What could be sources of load?
  • Has load increased? Has anything changed?
  • What do you have RAIDed? Looks like SCSI disks. 15k RPM?
  • Is the RAID1 your boot volume? USed for swap? etc

Without specifics it's tough to give advice, however I can state from experience that I had a RAID1 array which was a huge bottleneck on web/database server -- upgrading to RAID10 relieved the bottleneck. But I can't say for certain that this is your issue as well without more info.

EDIT: Based on your response that this is a cPanel server and the boot volume is RAIDed, you experience does sound similar to mine. I had a cPanel server with a few hunderd accounts and two SATA drives in a RAID1; I saw high load because multiple requests writing to those drives killed performance. RAID10 did the trick for me.

share|improve this answer
    
What does the server do? Web server? Database server? File Server? The server its a cPanel runing apache, mysql, exim, dovecot, pure-ftpd with 600 accounts. What could be sources of load? The Raid?! Has load increased? Has anything changed? Nothing What do you have RAIDed? Looks like SCSI disks. 15k RPM? WD 1TB 7200 Is the RAID1 your boot volume? Yes, all the partition are in raid1 /boot is /dev/md0 USed for swap? swap is in raid1 too, but never the system touch him –  melocoton Apr 20 '10 at 11:12
    
@melocoton: This would be easier to read if you edited your question and put it in there. –  Josh Apr 20 '10 at 13:27
    
@melocoton: Nevermind, I see that you did put it in the question. See my updated answer; consider RAID10. –  Josh Apr 20 '10 at 13:31
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.