|
On Mon, 24 May 1999, James Nessen wrote:
> I just read your post on the mon mailing list. I just had a couple
> of questions for you regarding the Linux Virtual Servers. I take it
> you are running the one listed in the HOWTO on sunsite. How is that
I'm not familiar with the HOWTO you speak of. We are using the vs-0.8
patch from the LVS Project at http://proxy.iinchina.net/~wensong/ippfvs/ .
> working for you guys? I just started a small web hosting outfit and I
> was thinking of beta testing the Virtual Services, so I would
> appreciate any feedback you could give.
LVS works fairly well for us but we're using a custom server not a web
server so YMMV. We've seemed to hit a performance bottleneck lately but
we haven't been able to discern whether it's a problem with our
non-optimal use of LVS or our hardware. You'll probably want to contact
the LVS mailing list to find someone with more experience with LVS as
we'll only been using it a few months.
LVS seems to work great until a server goes down (this is where mon comes
in). Here's a couple of things to keep in mind. If you're using the
Weighted Round-Robin scheduler, then LVS will still attempt to hit the
server once it goes down. If you're using the Least Connections
scheduler, then all new connections will be directed to the down server
because it has 0 connections. You'd think using mon would fix these
problem but not in all cases.
Adding mon to the LC setup didn't help matters much. I took one of three
servers out of the loop and waited for mon to drop the entry. That worked
great. When I started the server back up, mon added the entry. During
that time, the 2 running servers had gathered about 1000 connections
apiece. When the third server came back up, it immediately received all
of the new connections. It kept receiving all of the connections until it
had an equal number of connections with the other servers (which by this
time...a minute or so later...had fallen to ~700). By this time, the 3rd
server had been restarted after due to triggering a high load sensor also
monitoring the machine (a necessary evil or so I'm told). At this point,
I dropped back to using WRR as I could envision the cycle repeating itself
indefinitely.
Using mon with a WRR setup seems to be the safest best (for us anyways).
We still have the problem of an "unequal" number of connections between
the servers when a server goes down and comes back up. What we need is
some way to have mon change the weight on a server that is being added to
into the loop and then change that weight to match the other servers once
the number of connections is roughly equal. I think this would require
add another or modifying an existing scheduler though (I don't think you'd
want to handle this in userland).
Hope this helps.
Christopher
|
