[olug] Slow read speeds-SuSE

[Kevin D. Snodgrass]

--- On Wed, 4/28/10, Craig Wolf <CJWolf at mpsomaha.org> wrote:
> Ok, my backups from one of my linux
> servers takes 5 days to finish...my window is 36 hours, not
> working right.  Throughput of about 1.5GB/hour.  
> So in my testing, copying a 158MB directory to the server
> takes about 1 Minute 5 seconds.  Copying same structure
> off the box takes almost 7 minutes.  Network switches
> show no issues/error.  iostat shows slow to no
> throughput at all on read.  
>  
> SO, the million dollar question is, what should I look at
> for fixing this??  Do I use hdparm?
>  

I'm a little confused here, Craig.  First you say copying a 158MB file takes 1:05, then next sentence copying same file takes almost 7:00. ??? Should I assume the second is with the backup software?  If so, then there's your problem, either the backup software is crap or it is doing some compression or something that takes lots of time.

I would first do a process of elimination.  Try the simple stuff first.  Check hard drive access on each end:

=====================
# time dd if=/dev/zero of=delete.me bs=1024 count=1024
1024+0 records in
1024+0 records out

real    0m0.073s
user    0m0.001s
sys     0m0.009s
=====================

That run was on an old AMD Athlon 2800 (32-bit) with a WD 200GB IDE drive.

Do that on each machine with the output file being located "in the area of interest", i.e. same place as the file I/O will occur during the backup procedure.

Then, just to make sure, check read speeds:
====================
# time dd of=/dev/null if=delete.me bs=1024 count=1024
1024+0 records in
1024+0 records out

real    0m0.014s
user    0m0.000s
sys     0m0.004s
====================

(Notice I changed the if= and of= sequence!)

Then make sure copying files from [src] to [dest] works as expected.  [src] being the machine that will get backed up, [dest] being the backup archiver.  NFS mount a directory from [dest] onto [src].  Then just copy the previously created 1GB file.
===================
# time cp delete.me /mnt/[dest]/deleteme/

real    0m0.245s
user    0m0.000s
sys     0m0.006s
===================

100Mb/s ethernet with the builtin nForce2 ethernet controllers on both machines, basic 8 port 100Mb/s switch and no other traffic.

If these test all look normal*, then I would look at the backup software.  Check CPU usage on each machine during a backup attempt.


* Caution! Normal is a state I'm am only vaguely aware of even existing.

Kevin D. Snodgrass