|
I spent the rest of last week on this and I managed to get over it.
First I added 2 more real servers, one that is the same as the others
and one that was a lot more powerful. Then I realised that some of the
values I was changing were not being applied to the webserver, like
changing the max file handles. The other thing I noticed is that I had
That is rather bad. But I could have told you this if you gave me the
information I requested. Nevermind, you solved the issues.
the keepalived timeouts set way too low for doing tests like this. The
reason it was always hitting 1145 was that was about the time it was
taking to get to the time out in keepalived and then it was timing out
the real servers and dropping them from the table, then it brought them
Ahh, makes sense.
back so fast I did not notice. The program watch is great for this
(watch -n1 ipvsadm)
Yes, it even used to have a bug when invoked in realtime (-n -1) :).
With this and 4 real servers I was able to get over 108000 connections
through the load balancer to the real servers. At this point the client
machines started dying :)
In which ways? Load, file handles, memory, interrupt congestion?
ab blows chunks
Well, it does a bit more then your perl script does ;).
my kid sis can write perl scripts that are more useful for this
Yes, if you only need to get a TCP connection, you're right, once you
need to handle some of the html data, you might need to delve deeper
into perl programming.
all docs on lvs performance are out of date, to this end I need to write
a paper on this
Ok, what exactly is out of date on them? I'm simply asking so we can add
a note where appropriate.
keepalived rocks more than god (the company I work for funded some of
the changes we need as well)
I don't know God that well, but I share my opinions on the functionality
of keepalived and I'm happy to hear that your company funded Alex for
some code improvements!
If you have any further questions please feel free to email me. I will
post once I have done more benchmarking and written docs about it.
... I hope you will include the numbers I requested 3 times in your
document. Other than that we'll certainly appreciate some documentation
or paper on the performance of LVS.
Take care - RL
You too. Best regards,
Roberto Nibali, ratz
--
echo '[q]sa[ln0=aln256%Pln256/snlbx]sb3135071790101768542287578439snlbxq'|dc
| 