ISSUE: 001 NODE SYSTEM STABILITY / PERFORMANCE NOTICE
I've been able to track the cause of the hung VMs to resource exhaustion
caused by prolonged and high system I/O. This resource exhaustion is
caused by user applications. There is a reason most providers ban
certain activities -- it is because they cause problems. Users who run
these types of applications are clustered-together so that they don't
impact users who are trying to host websites or run forums. I have
personally witnessed the lights go solid on servers from all of the
activity taking place inside of the machine. In some instances, this
activity has exposed weaknesses on the hardware nodes such as disks with
bad-blocks because every square-inch of the hard-drives are being
written to or read from while the node is alive. Furthermore, I've been
able to identify a range of servers that require reboots to resolve
these issues and 99% of the time they are allocated to the high I/O and
high-bandwidth users of the service. Therefore, if you are one of these
users please keep in mind that your use and that of your peers will
most likely result in a server bounce every so often to correct
problems or you will experience performance that changes from moment-to-mement
this is something you will have to tolerate.
from: http://www.leeware.com/status.txt
Hmm I wonder who these high bandwith users are???
Reply With Quote
Originally Posted by haydeno
Bookmarks