vps network problem, machine hungs for 4-6 minutes

[dtorres]

My vps (at sedna.nocdirect.com) is hunging for 4-6minutes, 4-6 times per hour. It is mainly a development site, so ssh is the only traffic. Of course, ssh sessions die. When doing traceroutes from different providers, i've seen that connection between sedna and my vps is lost.

There are no memory issues (i've been watching cpu+memory during outages) and i'm using 220Mb out of 384Mb. load average is 0.0. Neither there are bandwith spikes. I'm using lighttpd+mysql, tuned to stay down on memory usage.

Also I was told about firewall issues, but trying traceroutes/pings from different providers shows the same results. Connection is lost from sedna to my vps.

At the end, working in this vps is a pain in the... and there is no way anyone could have a site that stalls for minutes. Maybe other vps in the same host is misbehaviouring? for 5 days???

Could anyone help? Looks like basic tech support is not able to find/fix the problem. (ticked id: 13055756) Last week everything was fine, but now it isn't usable....

[JPC-Awais]

dtorres,

We have updated your support ticket. Please check the reply and followup there.

[dtorres]

As I said before, I've recorded memory usage during machine problems. Having >164Mb of free memory isn't enough?

I've done tests with apache bench under heavy load, and webserver doesn't use more memory.

Shouldn't 384mb of ram be enough for ssh + webserve? (without control panel, bind, etc)?

[JPC-NickO]

Hi dtorres,

It really depends on what type of sites your running and the amount of concurrent connections you are receiving.

for standard apache+php it would be 15-20 concurrent connections would be pushing it without some serious optimization.

[dtorres]

Hi dtorres,

It really depends on what type of sites your running and the amount of concurrent connections you are receiving.

for standard apache+php it would be 15-20 concurrent connections would be pushing it without some serious optimization.

It is site that I use to develop. So maybe, with luck 1 concurrent connection. besides, lighttpd uses ten times less memory than apache.

however, if i'm hitting some kind of limit on the vps, it should be logged somewhere...

again, there is no problem with cpu load or memory usage. vps stalls, and it isn't a cfg issue. something is blocking connections between sedna.nocdirect.com and my vps, something that wasn't happening 10 days ago.

[JPC-NickO]

Hi dtorres,

You can type:

cat /proc/user_beancounters

Look in the failcnt column to see if your hitting any limits.

[dtorres]

Hi dtorres,

You can type:

cat /proc/user_beancounters

Look in the failcnt column to see if your hitting any limits.

vps is way below barrier/limits (as expected). so... can someone have a look at why machine freezes?

[JPC-Jawad]

Dtorres, I see your ticket updated. I would request you to follow up in the ticket regarding any support issue instead of public forum as this is not an official means of support.

[dtorres]

Dtorres, I see your ticket updated. I would request you to follow up in the ticket regarding any support issue instead of public forum as this is not an official means of support.

Totally agree.

This post was an alternative way to contact other people with similar problems... because, up to now, support hasn't been helpful in resolving this particular problem.