Linode intermittently unresponsive
Hello!
I was wondering if anyone might have some suggestions about how to deal with a Linode that is intermittently unresponsive.
I've been running a series of apps on Linode for several years and in the past I've always been very happy with my Linode performance. However, it seems that within the past year or so (since the Akamai transition?), I'm occasionally seeing brief periods of a few minutes where my Linode is completely unresponsive. When I look at the performance logs, I see long periods of flat, stable performance followed by a sudden spike where CPU, Disk IO, and Network all go up to near 100% for a few minutes before returning to the low baseline.
I was curious if others have seen similar performance issues and if anyone might have some thoughts as to what might be causing this or what a possible solution might be.
Thank you,
-abe.
3 Replies
Have you checked your firewall logs to see if there's any kind of bursts of network traffic?
-- sw
Hi. I haven't dug into the firewall logs yet. I'll try that. The network monitor shows usage pretty flat except for a burst this morning where it briefly went up to:
Public In: 6.14 Mb/sec
Public Out: 190 Kb/sec
Also the CPU load increased to about 60% over this short 5 minute period and then returned to the baseline.
This doesn't seem egregious but we were doing a demo during that time and the website became unresponsive at that time and then returned to normal after these interval, so the observed performance and the CPU/network spike seem correlated (aside from the demo gods playing games with us).
I have one other very basic question: On a VPS server, my server / application is a virtualized server hosted a piece of shared hardware with a bunch of others, so is it possible that performance variations might be caused by factors that are external to my particular server instance?
so is it possible that performance variations might be caused by factors that are external to my particular server instance?
A backup may have been running.
You may have experienced some kind of network brute force attack (the Chinese & Russians are famous for ssh brute force attacks)…that’s why logs are important (look at your auth.log & web server logs as well). A DDoS attack is probably unlikely…but you could consult support about this.
You may have suffered the side effects of “cycle steal”…one or more other VPSs getting more than their fair share of hardware resources (starving your VPS) for a short period of time. Consult with support about this.
— sw