On Fri, 23 Aug 2013, john boris wrote:
David
Thanks for the info. First I might be wrong on the actual number as this
server can have at most 24g of ram maybe 32 but it will not boot when you
put 16g in it.
As for the slow file transfer it is only one machine at a a time and
although I have 17 vms running only one is live and that runs fine and
small files transfer normally
hmm, then this probably isn't the network, but it could easily be disk I/O or
memory
Granted I have a lot to do and will do what you suggest as I need to get
this new server ordered within 2 weeks before my boss closes his wallet
these tests should actually be pretty quick. you should be able to get most of
the info within about 10 minutes (unless your apps are batched so that you would
need to test over a longer period)
by the way, you may also want to look at the output of vmstat to see if there is
anything in the si/so columns (indicating that it's using swap)
If you are using swap, try reducing the number of VMs on the hardware until you
aren't, so that you can see if there's something else you're probably going to
run out of (if you have to move half the machine off before you stop running out
of RAM, but this then shows you at 80% CPU, you are running out of CPU as well.
David Lang
On Friday, August 23, 2013, David Lang wrote:
On Fri, 23 Aug 2013, john boris wrote:
Okay I am doing this from my iPhone so I am trying to answer everything
1. I am tied to scp because my COBOL app needs a program called eroded
which will not port to Linux I have a lot of that covered but we are
moving
to PowerSchool in the cloud so I need these systems as is for historical
reasons
2. I have tested this with all 17 machines running on one of the hps and
although it works it is sluggish and file transfers suck
Ok, this is actually good news. It means that you have the ability to see
where your bottlenecks are and address them correctly with the new system
instead of just guessing.
A quick google search is telling me that the DL-160 G5 is
2 processor, 2-4 core (max 8 cores)
max 64G ram
2x 1G ethernet
4x hard drives (SATA or SAS)
how much ram and how many cores do you actually have?
unfortunantly a sluggish system and slow file transfers could be a lot of
things.
On the host OS, run 'iostat -xkt 60' or similar and look for a few things.
is there idle CPU or not?
is there a large amount of wait time?
do you have drives with amounts of I/O, especially ones that show
utilization up around 100%
Then look at memory stats (either from top, or cat /proc/meminfo)
are you using swap? how much free memory do you have (on a reasonably
freshly booted machine or shortly after a cache flush on the host OS)
You could just be running out of CPU, RAM, I/O, and/or network
setup something like MRTG to monitor your network I/O while the file
transfer is running, are your interfaces pegging at max throughput? or are
they much lower than the theoretical peak (if so, you will probably see
issues with CPU/RAM)
If you see lots of wait with no idle CPU, I would look first at RAM.
If you see lots of wait, but there is still idle CPU, I would suspect that
your disk I/O is your bottleneck (which iostat should show)
17 systems going through a single Gig-E interface could be a problem if
they are all trying to do file transfers at once, does your network support
10G? if it supports it, you should get it on your new server.
David Lang
_______________________________________________
Tech mailing list
Tech@lists.lopsa.org
https://lists.lopsa.org/cgi-bin/mailman/listinfo/tech
This list provided by the League of Professional System Administrators
http://lopsa.org/
