Unexpectedly, the FreeNAS 8.0.2 is very slow.

[bronzarr]

I used FreeNAS-8.0.2-RELEASE-amd64. On a daily basis I have 3 clients connected to the FreeNAS; Mac (mini) OS Lion and Apple TV iOS 4.2 over 100Mb Switch wire using AFP, and Windows7 x64 PC's over 100Mb Switch wire using smb.

I have used freenas for a month, it worked well, but yesterday when I turn off and turn on again, the device work very slowly.

when I play them movie by using Mac or Apple TV. The browse file work very slowly and When it is playing. it always stop to buffer every 5 second . It never be like this before.

On the server, in /var/log/messages, I have:

Code:

Nov 14 06:09:57 freenas afpd[2415]: AFP3.1 Login by bronze
Nov 14 06:10:02 freenas cnid_dbd[2424]: Set syslog logging to level: LOG_NOTE
Nov 14 06:10:30 freenas afpd[2415]: transmit: Request to dbd daemon (db_dir /mnt/Media2TB2) timed out.
Nov 14 06:10:30 freenas afpd[2415]: Reopen volume /mnt/Media2TB2 using in memory temporary CNID DB.
Nov 14 06:10:40 freenas cnid_dbd[2424]: read: Connection reset by peer
Nov 14 06:10:40 freenas cnid_dbd[2424]: error reading message header: Connection reset by peer
Nov 14 06:21:22 freenas cnid_dbd[2690]: Set syslog logging to level: LOG_NOTE
Nov 14 06:23:34 freenas afpd[2687]: sys_getextattr_size: error: Result too large
Nov 14 06:23:34 freenas afpd[2687]: sys_getextattr_size: error: Result too large

And I try to copy file by using window7.The speed is 512 KB/s. Basiclly, it should be at 10-20MB/s

when I test the speed by using iperf . it shows;

Code:

C:\>iperf -c 192.168.1.100 -t 60 -i 10 -f M
------------------------------------------------------------
Client connecting to 192.168.1.100, TCP port 5001
TCP window size: 0.01 MByte (default)
------------------------------------------------------------
[128] local 192.168.1.50 port 52691 connected with 192.168.1.100 port 5001
[ ID] Interval       Transfer     Bandwidth
[128]  0.0-10.0 sec   103 MBytes  10.3 MBytes/sec
[128] 10.0-20.0 sec   102 MBytes  10.2 MBytes/sec
[128] 20.0-30.0 sec   102 MBytes  10.2 MBytes/sec
[128] 30.0-40.0 sec   102 MBytes  10.2 MBytes/sec
[128] 40.0-50.0 sec   102 MBytes  10.2 MBytes/sec
[128] 50.0-60.0 sec   102 MBytes  10.2 MBytes/sec
[128]  0.0-60.0 sec   613 MBytes  10.2 MBytes/sec

Interface Traffic bring from Web GUI.

When I copy file from Mac between Win7 or playing movie from apple TV by sharing file from Mac or win7. it is not slow same as FreeNAS.

How this problem occur? And How can I fix this problem?

 

[ProtoSD]

Was there a lot of disk activity / were the LED lights flashing a lot?

Your system was probably doing the automatic monthly scrub.

 

[Milhouse]

Are your disks OK, what's the status of your pool - could it have become degraded? Check with "zpool status", will also tell you when the last scrub took place (or if it is still taking place).

 

[bronzarr]

Thank all of you for advise.

This is the result of check zpool status

Code:

[bronze@freenas] /> zpool status
  pool: Media2TB1
 state: ONLINE
 scrub: scrub completed after 4h14m with 0 errors on Mon Nov 14 10:21:36 2011
config:

        NAME                                          STATE     READ WRITE CKSUM
        Media2TB1                                     ONLINE       0     0     0
          gptid/89c95f11-f6fb-11e0-b0e1-e0699502cfd3  ONLINE       0     0     0

errors: No known data errors

  pool: Media2TB2
 state: ONLINE
 scrub: scrub completed after 4h13m with 0 errors on Mon Nov 14 10:20:47 2011
config:

        NAME                                          STATE     READ WRITE CKSUM
        Media2TB2                                     ONLINE       0     0     0
          gptid/a745bbd6-f6fb-11e0-b0e1-e0699502cfd3  ONLINE       0     0     0

errors: No known data errors

As a suggestion from above and the result from test a zpool status, I think the system is slowly came from the system will take a auto scrub. Because just moment ago, the system is coming in normally.

Please give me more detail;
- How do we know if the system doing the automatic scrub or not?
- How often should we do manual scrub?

 

[Milhouse]

Please give me more detail;
- How do we know if the system doing the automatic scrub or not?
- How often should we do manual scrub?

- Only way to tell is by checking with zpool status, and a calendar
- An automatic scrub will occur every 30 days since the last scrub. It is recommended that with consumer (rather than enterprise) drives, the data should be scrubbed every week

Though to be honest, with your current setup it's hardly worth scrubbing or even using ZFS at all. You should be using UFS instead.

You have no redundancy, and I'm not entirely sure what the scrub will do if it finds corrupt data, since it won't have any copies or parity with which to recover correct data from. With only single drives in each pool, that would explain very neatly why your performance dropped off a cliff during the scrub.