Hi,
I spent some time investigating the server, and here's an intermediate
result. I'm keeping notes here
http://wiki.lyx.org/Devel/TestServer
Generally I've noticed that sometimes even when interacting with the
server via CLI, it sometimes seems slow to respond. So I created a simple
script that executes a command while measuring the execution time.
Initially I wanted to look at disk access, but then I went with
something even more basic, e.g. the command
sleep 1
This command ought to take 1 second to execute, and if it takes much
longer than that something strange is happening, which could of course
just be a very high load on the server. Or perhaps on the host of the VM.
So once I had the script that measured these times, I gave it a threshold
time for which it'd count "overruns". Below is a table with results for
measuring the execution time of
sleep 0.1
with a threshold of 0.12 s, and doing this 10000 times. This test was
repeated on four different "machines".
Test Overrun Max VM/HW System
# [ms]
1 517 2600 VM LyX server (VM)
2 32 158 VM VM on Christian's home server
3 24 145 HW Christian's home server
4 2 174 HW Christian's old Mini-ITX
Test 1 is on the LyX server, and there are many "overruns". Sometimes it
takes one or even two seconds to execute 'sleep 0.1'. This is _much_ more
compared to test 2-4, which was run on other hardware or virtual machines.
As for conclusions, I know think I have a test that gives a difference for
the LyX server compared to other machines. I think the most telling sign
is when a command such as 'sleep 0.1' takes over a second to execute, then
something strange is going on with the server.
/Christian
--
Christian Ridderström Mobile: +46-70 687 39 44
