I know that there a already a lot of threads about the slowness issues, but i was not able to find an answer. Here is the issue:
FTP read from NAS is about 30-70MB/sec - which is fine to me
FTP write to NAS - 10-12MB/sec - slow!
CIFS read from NAS - 12MB/sec - slow!
CIFS write to NAS - about 24-30MB/sec - little bit slow, but i can live with that
So what is weird is that FTP write and CIFP read are slow. I'd expect for read for both or write for both to be slow, but not like this.
Some info about my system:
Core 2 Duo - top shows 3-5% when i copy
4GB Ram - still have a lot of mem when copy
4x2GB WD Green Drives in ZFS
speeds:
> dd if=/dev/zero of=/mnt/force/zerofile.000 bs=1M count=4000
4000+0 records in
4000+0 records out
4194304000 bytes transferred in 118.201466 secs (35484365 bytes/sec)
> dd of=/dev/null if=/mnt/force/zerofile.000 bs=1M count=4000
4000+0 records in
4000+0 records out
4194304000 bytes transferred in 23.113810 secs (181463118 bytes/sec)
network
ifconfig | grep media
media: Ethernet 1000baseT <full-duplex>
media: Ethernet autoselect (none)
i have some intel NIS and was trying to do this:
media 1000baseTX mediaopt full-duplex
makes no difference
as well as
AIO, sendfile(2) and Large RW support (all of them are enabled now)
Please help! I want to understand what is happening and how to fix this.
Bob
BTW, for developers - it would be really nice to have some tool that checks where is the bottleneck or at least cheatsheet. It looks like this is one of the common questions on a forum.
FTP read from NAS is about 30-70MB/sec - which is fine to me
FTP write to NAS - 10-12MB/sec - slow!
CIFS read from NAS - 12MB/sec - slow!
CIFS write to NAS - about 24-30MB/sec - little bit slow, but i can live with that
So what is weird is that FTP write and CIFP read are slow. I'd expect for read for both or write for both to be slow, but not like this.
Some info about my system:
Core 2 Duo - top shows 3-5% when i copy
4GB Ram - still have a lot of mem when copy
4x2GB WD Green Drives in ZFS
speeds:
> dd if=/dev/zero of=/mnt/force/zerofile.000 bs=1M count=4000
4000+0 records in
4000+0 records out
4194304000 bytes transferred in 118.201466 secs (35484365 bytes/sec)
> dd of=/dev/null if=/mnt/force/zerofile.000 bs=1M count=4000
4000+0 records in
4000+0 records out
4194304000 bytes transferred in 23.113810 secs (181463118 bytes/sec)
network
ifconfig | grep media
media: Ethernet 1000baseT <full-duplex>
media: Ethernet autoselect (none)
i have some intel NIS and was trying to do this:
media 1000baseTX mediaopt full-duplex
makes no difference
as well as
AIO, sendfile(2) and Large RW support (all of them are enabled now)
Please help! I want to understand what is happening and how to fix this.
Bob
BTW, for developers - it would be really nice to have some tool that checks where is the bottleneck or at least cheatsheet. It looks like this is one of the common questions on a forum.