Red Hat Bugzilla – Bug 168969
Nfs server very slow
Last modified: 2007-11-30 17:07:20 EST
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.10) Gecko/20050909 Fedora/1.0.6-1.2.fc4 Firefox/1.0.6
Description of problem:
I export a raid directory with nfs. I mount the share on the local machine or an other host. I copy 160Mb in about 90s. a normal value is 15..30s. There is no error log somewhere. If I export in async mode It copy in 5s + 12s for the sync command on the server so it should take about 17s and not 90s. Same result without raid (just a disk).
I tested with a debian and kernel 126.96.36.199 and 2.4 it's the same. Debian also provide a user space nfs server. It's then take about 18s.
I tried to update to your last kernel but then I have a kernel panic.
Local copy on the exported directory take about "time cp -av /opt/usr /mnt/sdb1/usr5 ; time sync" 15s
Installation on a dual Intel(R) Xeon(TM) CPU 3.40GHz with a RAID bus controller: LSI Logic / Symbios Logic MegaRAID (rev 07). all info for hardware at https://rhn.redhat.com/network/systems/details/hardware.pxt?sid=1005916035
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1.echo "/mnt/sdb1 xxx.xx.xx(rw,no_root_squash,sync,insecure)" >/etc/exports
2.service nfs restart
3.mount xxx.xx.xx:/mnt/sdb1 /mnt/test
4.time cp -av /opt/160MB-dir /mnt/test
Actual Results: time is about 90s
Expected Results: about 18s
I can not update to the latest kernel as the kernel panic
Problem solved. The scsi disks are incompatible with the back-plane.
I used 5 Seagate disk and all of them add problem. With Maxtor disk it work fine.
It's a strange behavior as the disk are fine under Windows (all test done). It's
also not a problem with the back plane as I tested with other back plane (4
diff.). It's also not an raid card problem as i changed to std 29160 and the
problem was not solved.