jonathan
By:
jonathan

Amsterdam server processing speed seems very slow but bandwidth is fine.

June 18, 2013 6.1k views
Doing anything like apt-get update seems to be taking an absolute age. I did a wget -O /dev/null - http://cachefly.cachefly.net/100mb.test and got 30mb/sec, and average speeds for apt-get downloads seemed about 12mb/sec so the problem isn't there. But even things like htop, which is just a slightly more advance "top" show as taking 10% of my CPU allowance. It takes a good 5 seconds to show the "welcome" message, when my previous non-SSD basic VPS was completely instant (for comparison, htop either doesn't show at all, or takes just 1% CPU on the other VPS). My details are 512MB Ram 20GB SSD Disk Amsterdam 1 Ubuntu 13.04 x64 Server In the admin setup log, it says it took 115 seconds to provision; as this is 2x as long as it suggests it should take, I'm wondering if there are some issues with that machine or region? BTW, I rebooted the machine before I started typing this, after giving it some virtual memory (as required by the Dropbox linux client) same as I have with other VPS, and 5 minutes later I'm still getting "connection refused". I'd normally expect a reboot to take around 20 seconds on a VPS.
4 Answers
Update - thankfully I found the Virtual Console (shame you can't copy/paste) but basically it seemed to be "stuck" at

"Killing all remaining processes ... [fail].
"Will now switch to single-user mode.
Give root password for maintenance or type Control-D to continue".

Giving root password just brought me back to the root@www prompt.

Entering "reboot now" just repeats the "Killing all remaining processes ... [fail]" message.

Had a google and found this: http://ubuntuforums.org/showthread.php?t=2139160

So basically, "reboot now" shouldn't be used with 13.04. I'll post another thread to help others!

So I've got my machine back but it seems now quicker. I fired up a new droplet in NYC and it's LOADS quick but as I'm in Europe having AMS working at full speed would be preferable.
Second update: Looking at twitter, it seems other are having problems with Amsterdam for at least 5 days now. DO twitter support say to open Amsterdam slowness problem tickets with "urgent" status, so that's what I've just done. Hoping that I can configure my NYC droplet and then move to AMS once fixed...
Correct @jonathan. We recommend opening up a support ticket as it will help up determine the issue better in this case.
OK, I have an answer from support and it worked for me! If you find you are on a slow server in the Amsterdam region, here is their suggestion:

---------------------------------------
"We do not have an ETA to fix, however there is a workaround.

You may simply make a snapshot of your droplet. Then try to re-create a new droplet from that snapshot. Once it's up and you login, you can do a "cat /proc/cpuinfo" if you do not see "Intel" there, your droplet maybe buggy.

You can repeat this a few times, while keeping the old droplets active, and you should get an Intel box."
---------------------------------------

To clarify, this means that you don't simply restore your droplet, but on the "new droplet" page, you choose a size and region, and then under "images" you choose the tab called "my images", select the backed up image, and then just choose "create droplet". I knew things were going better when it took just 45 seconds to create rather than the previous time of almost 3 minutes, and logging in was instance and it was running like a dream.

cat /proc/cpuinfo showed that I was on an Intel box with 8x the cache of the AMD box.

So, there's the solution. Of course, if you haven't done any configuration yet and you've got a slow box, just kill the "bad" droplet and try again.

Hope this helps someone!
Have another answer? Share your knowledge.