Frustrations on VPS

[byteme]

I am somewhat concerned that having seen the mailserver smtp that is used by my VPS suddenly shut off and refuse more mail, an answer is not forthcoming.

I have noticed at the same time, the server CPU turns red and the load increases by some 65 - 70%.

I have usually been running or attempting to run a cron job (sometimes 2 on different domains) that sends out mail 10 per min and 2 per min (one domain sends at 10 pmin and another sends at 2 pmin)

What then happens is I check the logs to discover the CRON jobs are NOT sucking the load of the CPU e.g. %CPU 0.1

I did see %CPU 7.2 /usr/sbin/exim -bd -q30m which seems to suggest loads on the mailserver I think.

Anyway, I saw the failed cron (it emails me) I checked the WHM and saw the CPU load so I entered a chat session with VPS support.

- I kept a copy of the logs.......

I said : Server Load 2.74 (2 cpus) Memory Used 62 % Swap Used 23.27 % Disk /dev/vzfs (/) 35 %

After I explain they look and say:-
please place a ticket for the issue and we will update you soon in the ticket

I state
they then agree to look.

To cut a long story short, they tell me it is now green

I reply: I know it is green right now, but you seem to miss the point, the cpu periodically has a fit. It goes red on my whm and reads 2.7 or whatever. When this happens it stops the cron process in its tracks by halting the mail smtp

them: well i will suggest you to let us knwo when yoiu are in the middle of the problem

them: that will help us to understand the whole issue

This is the third time (or 4th?) that this has happened and still no closer to the issue.

Tried a ticket, no help, tried chat been also told that I should monitor my VPS to see why (!!!) If I could discover why SURELY I wouldn't need support.

Anyone ? What do i have to do to get a proper answer based on fact?

[Administrator]

Hello Ros, Performance of our old x3210 nodes has gone down in last couple of months and we are migrating all our customers on new Dual Quad Core Dell servers which should solve these problems. Please make it convenient to submit a ticket for our VPS department and get them to schedule migration of your VPS on our new node. They will take care of migration process and we will ensure that you don't face any sort of downtime in this process.

[byteme]

Well here we are 6 weeks later with my new VPS. To say it went smoothly, well it was better than the last move.

Thing is I am still having the same problems of the CPUS going into Red and flashing on WHM. I was informed WHM is sick on the status side (??) and to look at virtuozzo.

Virtuozzo shows completely normal EXCEPT...............
during the red flashing blip on the WHM the server is slow to allow FTP or mail or log in. Still shows websites most of the time.

These blips are only for a few mins, but tracking why seems to be a mystery.

Just how do I work out what is going on here??? Support always looks, says give some time and then comes back after 5 mins saying all is fine yes at that point it is, but I want to know why it goes blip! How can we find out what is going on??

[Administrator]

Do you know if such problems occur during any specific period of the day ?

I need some clue to assign someone to monitor your VPS node as we need detailed information to pin-point such problems.

[byteme]

Dear Admin,

I am not sure if it is always the same but usually the morning is the time it takes place around 9.00 to 9.30 ish, but I will see if I can catch exactly when over the next few days if you like so that it will help the monitoring of the VPS.

Thank you, it is driving me nuts!

[Administrator]

Originally Posted by byteme

Dear Admin,

I am not sure if it is always the same but usually the morning is the time it takes place around 9.00 to 9.30 ish, but I will see if I can catch exactly when over the next few days if you like so that it will help the monitoring of the VPS.

Thank you, it is driving me nuts!

I understand how frustrating it is. there might be someone else who may have some sort of cron set which runs at a specific time to overload whole node. We need to catch him and thats why we need some detailed information from your side.

[byteme]

will check it each day and see if I can verify days and times