Terrible typing lag on 1GB Droplet

April 24, 2014 5.4k views
Is there some sharing algorithm on Digital Ocean that locks up sessions for seconds at a time while logged in over SSH? Here's my experience: I'm logged in to my 1GB RAM Droplet (Ubuntu 14.04) and I'm setting up services, editing files with vim, etc. I can type in bursts with no lag for maybe 20-30 seconds, then it is completely frozen (not responsive to keystrokes) for 3-5 seconds, then anything I typed while frozen gets all sent at once. This is extremely annoying, especially if I'm using the backspace key... Is this a know issue or a temporary problem? It doesn't happen to me with Linode. Thanks
7 Answers
It's a latency problem as far as I can tell. I get it from time to time. My droplets are in the NY2 data center, and I’m on the west coast. I don't know exactly where the holdup is.
I am a new user as of 24 hours ago. I created three droplets, two in Amsterdam and one in NY2. I am in London.

The two in Amsterdam are great, no disconnects, no lag, very nice.

The one in NY2 has bad ssh typing lag but in my experience it's always while waiting for disk (e.g. tab completion, trying to run something, etc.) as if there is no i/o available.

My ssh sessions to NY2 also randomly disconnect which may or may not related.

Something broken in NY2?
I had the same issue for San Francisco and I'm in Seattle. Not too far away. So I moved it over to NY2 but it is equally slow.

I can power cycle remotely from the site and then, and only then, does my site work...for about 20 minutes. Then it because staggeringly slow and unresponsive, through browser and ssh-ing.

Is NY1 the best one...?
Just had this happen on my VM.

Signed up last night and i'm finding that there is terrible lag out of NY2 for some reason. I just run a simple game server (it only requires 128mb at recommended..) and the thing lagged hard, timing everyone.

I ssh in, kill the process (to be safe), check the uptime.. 0.00 0.13 0.10, then.. WHAM, laaagggggg on the ssh session to the point I can't type. I could not even ping the server while this was happening. Google pinged fine :-/

Same exact issue here with their San Francisco DC and I live less than 50 miles away. I created a new droplet in hopes it would get placed on a different node, and the new droplet was fast at first (like an hour tops) and then it was slow and unresponsive again. I can't work on my server like this, everything takes too long and it's beyond frustrating when I can't even type anything.

Doing a traceroute to the droplet IP showed no hops that were slow or any bottlenecks, so I doubt the issue has anything to do with latency (I get 28ms to the server) or shitty carriers... as mike mentioned above rebooting the droplet does seem to fix the issue temporarily (logging in is fast, typing is fast, everything is fast...)

Not sure why it goes from being fast to terribly slow, especially on fresh droplets that I haven't even done anything with.

I'm having the same problem. Horrible typing lag over SSH to NY. I just spun up the smallest slice because of the FreeBSD addition. It's almost unusable; my linodes never have this issue regardless of where the linode is in the USA versus me. I think it lags worse than doing dyndns with my home internet. My pings are in the 50-80ms which should be PLENTY fine for simple SSH command line stuff.

Any ideas why?

Same problem here. Too much lag in SSH. I've created the droplet in Singapore and I'm living in India. I also tested AWS instance in Singapore which works awesome without any lag. Please look into this.

  • I'm having this same problem. 3 droplets in Singapore. I'm located in Malaysia.
    All 3 droplets have very few traffic, and low CPU/RAM usage.

Have another answer? Share your knowledge.