|
Horms wrote:
Hi,
I am looking for information from people who are running LVS in
production. I am most interested in how reliable they find
LVS to be. How often does the system go down. And when it does
what is the cause - if it is caused by something unrelated
to LVS like faulty hardware then you can skip that bit.
hm .. we 've a setup here at work w/ lvs-nat on two ibm x305 boxen
running lvs-1.1.7 currently on linux 2.6.0-test5 (debian/woody)
We have ~30 VIPs on six gigE interfaces .. traffic accross all
interfaces averages at 15000 packets/s .. there is also an
iproute2 setup in place.
so far we've encountered no stability issues wrt/ lvs, its
rock solid, even the (carefully selected ;-) devel-versions.
the only problems so far were ..
a) constantly hitting connection-tracker limits
b) connection-syncronization with the second node does not seem
to work in our current setup (results in kernel-oopses)
c) mon is not exactly "intelligent" when it comes
to monitoring hundreds of services .. (it then puts quite
some (artificial?) load on the director)
and a subjective observation
d) if the setup reaches a certain level of complexity (especially
w/ the iproute2 stuff) .. debugging and troubleshooting becomes
a "non-trivial" task .. and it then sometimes can happen
that even ten eyes overlook a typo in the config ;-)
otherwise lvs rocks; kudos to the developers :-)
regards,
alex.
| 