Sunfire issues...

smartquant · #1 02-07-2007, 08:49 AM

Hello,

Is anyone else having problems on Sunfire?

Loading times have been slow since yesterday...especially on those sites that are using MySQL databases.

riostyles · #2 02-07-2007, 12:34 PM

Yes about since a week, maybe 10 times down but always about 30 mins.

Serra · #3 02-07-2007, 01:46 PM

If you've been down that much, ask for a refund for this month!

smartquant · #4 02-07-2007, 02:18 PM

Thanks for the input.

Sunfire has been very shaky lately.

I sent in a support ticket and the tech replied:

"Hello,
Currently there are no issues to report on this server."

When I received his email things seemed to be OK. Currently at 3:14EST
I'm having challenges again.

Serra, thanks for the suggestion. Have you presonally requested and received this type of refund before?

Thanks again to you guys for your input.

riostyles · #5 02-07-2007, 04:43 PM

Yes I was checking for the discount but to much work for a few bucks and at the other hand we had about 9 months wit very little down times, maybe 30 mins in 9 months.

There is probably a reason that HG will keep secret like an attack or an inner cript misshandle/use because I observed that the load/memory/even swap goes up to the top before the crash.
I had some differents answers to my tickets, depends of the admin "Level" answering.
Remember when they had the virus on an other server, ... similar.
Also I remember me some other servers in the past where you do not had any answer at all an some times days and days down.

Serra · #6 02-07-2007, 06:20 PM

Quote:

Originally Posted by smartquant

Serra, thanks for the suggestion. Have you presonally requested and received this type of refund before?

No, I have not personally, but many have, and they do get refunds. The key is documentation. (3rd party monitoring will not count as documentation)

smartquant · #7 02-07-2007, 08:03 PM

Quote:

Originally Posted by Serra

No, I have not personally, but many have, and they do get refunds. The key is documentation. (3rd party monitoring will not count as documentation)

I see. Thanks for your answer. I appreciate that.

Thanks to everyone for their input.

Serra · #8 02-07-2007, 09:10 PM

I just want to point you to this post, which was just posted:

http://forums.hostgator.com/showthre...6591#post46591

Someone basically using their party monitoring to watch their site and now claiming downtime of greater than 99.98% wanting a refund. Not going to happen, you can see HG's response.

When you ask for a refund, don't quote 3rd party stats, you'll never get a refund. Quote ticket numbers and contact times, times you noticed downtime and anything you can get them to admit to. They will refund your money if you have documentation, but contacting support and logging tickets is the ONLY proof they will accept. Not 3rd party monitoring... don't even talk about it...

HostGator doesn't like to give away refund/credits for downtime. Its that simple. If you don't do your homework, you will never get a refund. If you are due a refund and have tickets to show you worked with Hostgator to monitor and document downtime, they will give you refunds (or more likely credits).

smartquant · #9 02-08-2007, 09:56 AM

Thanks for that info, Serra.

That's good to know.

By the way the loads on Sunfire have improved and are showing green lights now.

The only yellow light is showing on the Disk sda3 (/usr) 90%.

I don't know what that means but I believe a yellow light is a sign meaning either "warning", "caution", or something like that.

Does anyone know if a reading of that type might slow sites down or make them harder to access?

Impavidus · #10 02-08-2007, 10:07 AM

/usr contains read-only data, and it unlikly causing any server issues. The amber color light means caution.

smartquant · #11 02-08-2007, 10:15 AM

Thanks for that info, Impavidus. Much appreciated.

Serra · #12 02-08-2007, 10:26 AM

Quote:

Originally Posted by smartquant

Thanks for that info, Impavidus. Much appreciated.

On that disk, it will likely go red before they even look at it. The only disks that you need to worry about are /home and /tmp. If either of those ever go red, contact support.

smartquant · #13 02-08-2007, 10:35 AM

Quote:

Originally Posted by Serra

On that disk, it will likely go red before they even look at it. The only disks that you need to worry about are /home and /tmp. If either of those ever go red, contact support.

Hello Serra,

That's good to know.

I really appreciate everyone's input.

Thanks.

riostyles · #14 02-09-2007, 06:01 AM

Again down

smartquant · #15 02-11-2007, 02:47 PM

Hey riostyles,

It looks like we are up. The only red light I see in service status is the following:

Disk sda3 (/usr) 91 %

I usually it's yellow and I can't remember the last time it was green.

I read earlier in this thread that the only disks that you need to worry about are /home and /tmp.

riostyles · #16 02-11-2007, 04:04 PM

So long the usr disk is not full, this make no problem at all (As I know)

riostyles · #17 02-11-2007, 08:57 PM

Hi smarquant,
I'm not a nix guy and we had always this concern even on other servers.
In the past I observed that the filling was higher end of the months then was down about the 2nd of the month. I interpretated thats some tmp files there like stats before compresssion but....
Looking around google and some webmaster forums everybody seems to have the same concern -- without troubles.
This settings seems part of the cPanel HD partitioning.
Voila, I'm not in trouble with this.

smartquant · #18 02-13-2007, 04:52 PM

Hey riostyles,

Thanks for that info.

I just put another support ticket in today. Here's what the server status looked like. Some sites running slowly or not at all.

Server Load 30.74 (4 cpus)
Memory Used 89.4 %
Swap Used 100.00 %
Disk sda1 (/) 25 %
Disk sda7 (/home) 67 %
Disk sda6 (/tmp) 3 %
Disk sda3 (/usr) 92 %
Disk sda2 (/var) 40 %

riostyles · #19 02-13-2007, 06:23 PM

I had even a crazy 120 load

Serra · #20 02-13-2007, 10:10 PM

Quote:

Originally Posted by riostyles

So long the usr disk is not full, this make no problem at all (As I know)

Yes, correct. The only disk that is trouble is /tmp and /home, so /usr can be red and the system will still work fine.

Quote:

Server Load 30.74 (4 cpus)
Memory Used 89.4 %
Swap Used 100.00 %
Disk sda1 (/) 25 %
Disk sda7 (/home) 67 %
Disk sda6 (/tmp) 3 %
Disk sda3 (/usr) 92 %
Disk sda2 (/var) 40 %

When you see loads that are more than 2X the # of CPU and Swap going to 100%, its a sure sign of one of two things, either the box is being attacked externally or a user is running an out of control script.

Normally, if it is a user, that will last until support zaps them. Then that user will spend 4 or 5 days talking to abuse before they are allowed back. If the problem doesn't clear up shortly, you can assume it is an external attack, which doesn't get much discussion here because that would be a bad idea. (Let them know what they are doing is working).

Quote:

Originally Posted by riostyles

I had even a crazy 120 load

A 120 load is totally insane, the box will not run for long on a 120 load. It will soon crash.

One of the few things that HostGator does well is handle over loaded servers. They are truly experts at it. You can guess that if your machine is running at 120 load, it isn't going unnoticed by HG.

kmaw · #21 02-13-2007, 10:20 PM

So, at 120/4 cpus is 30... I believe that's how that gets figured out... that's still nuts... 20 is usually the highest I see on my POS celeron when I tax it to the limit...

riostyles · #22 02-16-2007, 04:46 PM

Yes the 120 was a 120/4.
Down again right now. Yesterday too for a 30'
Having many down time since Jan. 30th, after about one year brilliant,
I set-up some monitoring 10 days ago.
I see the load pretty stable between 1 and 3 /4 cpu(s).
But the memory use, twice a day goes up to 90%, both and the swap up's first.
Notice that, as Serra said earlier, HG manage well the system.
Of course when we come to 0 swap free and 100 k mem free for 4Giga installed the load hurrys up and many times HG (I believe) cleans up before the crash, some times they are not fast enough and we dive.

Somethimes I alerted HG but they told me (Sollicitated answer for not to ad pain to the pain) that they have a kind of general status control panel.
Now since 2 weeks of trouble I believe this is not some fanzy script.
Of course if its some kind of DOS or memory overflow attack we will not here about at least before resolved and more probably never.
Some opimion?

riostyles · #23 02-18-2007, 11:31 AM

Right now, to late and we crashed

***GatorFord*** · #24 02-18-2007, 02:13 PM

My apologies, I'm looking into the users on the server to find out what is causing the high loads and outages. Thanks for bringing things like these to our attention. It will get sorted out.

riostyles · #25 02-18-2007, 03:09 PM

Hosting since about 15 years, you guys are very good.
When the trouble is killed/fixed, please advise us so we can back on work as usual and do not check our blackberry evey 5 minutes.