Web Hosting Talk : Web Hosting Main Forums : Industry Announcements : Providers and Network Outages and Updates : Semoweb Up And Down Constantly

[Erawan Arif Nugroho]

Um.. So you have the 10TB bandwidth VPS package. I was having that too, but I just cancel it a few days ago, because it's not stable.
I have the same issues as you. Sometimes it shows offline, and suddenly rebooted.

[moltra]

I too have the 10TBVPS package and have already had tech support move me to a new node due to random stability problems. The new node looked good at first, but now it is starting to randomly be off line. Tech supports asks for proof of a problem, when all they have to do is look at the statistics for the VPS.

I have noticed alot of people in this thread saying that the way the data is being collected could be the problem. The data I have is coming straight from Semoweb's VPS Control Panel by SolusVM.

If you look at the charts I saved from the control panel you can see where the memory used is zero. This means that their VPS monitoring system is logging that the VPS is down. So why do they need more information from the end user, if they can tell that the VPS is down.

And as you can see from the log, I have not rebooted or shutdown the VPS since 5/3/2012.

Judging by the responses to the tickets, Tech support is checking to see if the VPS is up, not what is causing it to be unstable. With multiple people complaining about the same problem, I would look at the hosting not the end users.

[Flapadar]

Originally Posted by moltra

I too have the 10TBVPS package and have already had tech support move me to a new node due to random stability problems. The new node looked good at first, but now it is starting to randomly be off line. Tech supports asks for proof of a problem, when all they have to do is look at the statistics for the VPS.

I have noticed alot of people in this thread saying that the way the data is being collected could be the problem. The data I have is coming straight from Semoweb's VPS Control Panel by SolusVM.

If you look at the charts I saved from the control panel you can see where the memory used is zero. This means that their VPS monitoring system is logging that the VPS is down. So why do they need more information from the end user, if they can tell that the VPS is down.

And as you can see from the log, I have not rebooted or shutdown the VPS since 5/3/2012.

Judging by the responses to the tickets, Tech support is checking to see if the VPS is up, not what is causing it to be unstable. With multiple people complaining about the same problem, I would look at the hosting not the end users.

While those graphs can be indicative of problems, that can also occur without one being there. You should collect further evidence to provide them with - for example, MTR reports.

[moltra]

How would a MTR report which is a network diagnostic tool, give more information than the hosting providers own, Virtual Private server management system that is used to provision the VPS? I can see the MTR giving information on the system being unreachable, but not with the load and memory usage being zero, which is an indication of the VPS being shutdown.

Here is a TRACEROUTE from the 1st node I was on. It shows a problem past the Hostdime.com. This was sent to semoweb with one of the earlier tickets I submitted to them.

Run 1
Traceroute #1 to 192.168.102.1
Hop Time (ms) IP Address FQDN
1 0.688 192.168.102.1 192.168.102.1
2 4.643 96.255.249.1 L300.WASHDC-VFTTP-85.verizon-gni.net
3 6.882 130.81.180.208 G0-3-5-6.WASHDC-LCR-21.verizon-gni.net
4 9.885 130.81.199.30 so-7-1-0-0.RES-BB-RTR1.verizon-gni.net
5 7.278 152.63.34.21 0.ae1.BR2.IAD8.ALTER.NET
6 26.908 4.68.62.137 ae17.edge1.washingtondc12.level3.net
7 32.011 4.69.158.18 vl-3501-ve-115.ebr1.Washington12.Level3.net
8 22.447 4.69.148.105 ae-6-6.ebr1.Atlanta2.Level3.net
9 31.646 4.69.137.149 ae-1-8.bar1.Orlando1.Level3.net
10 32.446 67.30.140.2 HostDime.com-10G-ethernet.core1.level3.net
11
12
13
14
Run 2
Traceroute #2 to 192.168.102.1
Hop Time (ms) IP Address FQDN
1 0.453 192.168.102.1 192.168.102.1
2 4.975 96.255.249.1 L300.WASHDC-VFTTP-85.verizon-gni.net
3 7.186 130.81.180.208 G0-3-5-6.WASHDC-LCR-21.verizon-gni.net
4 7.269 130.81.199.30 so-7-1-0-0.RES-BB-RTR1.verizon-gni.net
5 7.901 152.63.34.21 0.ae1.BR2.IAD8.ALTER.NET
6 37.013 4.68.62.137 ae17.edge1.washingtondc12.level3.net
7 17.467 4.69.158.18 vl-3501-ve-115.ebr1.Washington12.Level3.net
8 32.237 4.69.148.105 ae-6-6.ebr1.Atlanta2.Level3.net
9 29.755 4.69.137.149 ae-1-8.bar1.Orlando1.Level3.net
10 32.349 67.30.140.2 HostDime.com-10G-ethernet.core1.level3.net
11 32.846 184.171.255.59 mail-out.dixhost.com
12 34.137 184.171.255.136 184-171-255-136.static.dimenoc.com
Run 3
Traceroute #3 to 192.168.102.1
Hop Time (ms) IP Address FQDN
1 0.508 192.168.102.1 192.168.102.1
2 4.778 96.255.249.1 L300.WASHDC-VFTTP-85.verizon-gni.net
3 7.246 130.81.180.208 G0-3-5-6.WASHDC-LCR-21.verizon-gni.net
4 87.336 130.81.199.30 so-7-1-0-0.RES-BB-RTR1.verizon-gni.net
5 9.976 152.63.32.141 0.ae1.BR1.IAD8.ALTER.NET
6 36.929 4.68.62.133 ae16.edge1.washingtondc12.level3.net
7 19.985 4.69.158.18 vl-3501-ve-115.ebr1.Washington12.Level3.net
8 24.850 4.69.148.105 ae-6-6.ebr1.Atlanta2.Level3.net
9 42.196 4.69.137.149 ae-1-8.bar1.Orlando1.Level3.net
10 49.685 67.30.140.2 HostDime.com-10G-ethernet.core1.level3.net
11 34.942 184.171.255.59 mail-out.dixhost.com
12 37.285 184.171.255.136 184-171-255-136.static.dimenoc.com

Originally Posted by Flapadar

While those graphs can be indicative of problems, that can also occur without one being there. You should collect further evidence to provide them with - for example, MTR reports.

How could those graphs occur without there being a problem? If you are saying their monitoring system cannot enable them to tell if they have a problem, then they should look at another. When a server is suppose to be up, the memory usage should never be zero.

[Flapadar]

Originally Posted by moltra

How would a MTR report which is a network diagnostic tool, give more information than the hosting providers own, Virtual Private server management system that is used to provision the VPS? I can see the MTR giving information on the system being unreachable, but not with the load and memory usage being zero, which is an indication of the VPS being shutdown.

MTR shows where there is a connectivity problem. If their nodes were physically going down, I'm sure they would have noticed by now.

How could those graphs occur without there being a problem? If you are saying their monitoring system cannot enable them to tell if they have a problem, then they should look at another. When a server is suppose to be up, the memory usage should never be zero.

SolusVM is very buggy - for example quite frequently it shows clients using 60PB/s of bandwidth. Nor is it a monitoring tool.

[moltra]

I have a question for you. You get home and find out that your electric is out. What do you do? Do you call your electric company and report it? Or do you get in your car and start following the power lines to find out where the power is out? I can see giving the company information that might help them, but is that information useful, when their own systems are telling you that they are offline.

I know if I was running a company that supplied a service to consumers, (which is what my website is suppose to do), then I would make sure that I had the required infrastructure in place to ensure that the service was monitored and I was informed immediately if there was a problem.

For example, the data in my site is currently being used by one client and three more in the developement stage, I monitor my website and API constantly to ensure that my clients are getting what I promised them. I use various methods to ensure that my website and database is aviliable 24/7. And the service that I provide is at no cost to the clients.

Originally Posted by Flapadar

MTR shows where there is a connectivity problem. If their nodes were physically going down, I'm sure they would have noticed by now.



SolusVM is very buggy - for example quite frequently it shows clients using 60PB/s of bandwidth. Nor is it a monitoring tool.

So you are saying that they are using a subpar Virtual management system. If they are using a buggy virtual servar manager, then are they serving their clients? I know I would want as rock solid management system as possible. If the one I was currently using was "very buggy", then I would find another system, or put in place a method of telling if part of my system was down or unavailable.

[moltra]

@Flapadar, what would be a good site to use to monitor connectivity and help determine the cause of the reliability issues? I am not here to put down a company, or jsut to gripe. I have setup a generic test on pingdom to check the site, and I am looking for a site that does a traceroute on a regular basis.

[Flapadar]

Originally Posted by moltra

So you are saying that they are using a subpar Virtual management system. If they are using a buggy virtual servar manager, then are they serving their clients? I know I would want as rock solid management system as possible. If the one I was currently using was "very buggy", then I would find another system, or put in place a method of telling if part of my system was down or unavailable.

As a management system it is fine; but the statistics leave a lot to be desired. Take them with a pinch of salt.

Originally Posted by moltra

@Flapadar, what would be a good site to use to monitor connectivity and help determine the cause of the reliability issues? I am not here to put down a company, or jsut to gripe. I have setup a generic test on pingdom to check the site, and I am looking for a site that does a traceroute on a regular basis.

Pingdom for network uptime monitoring, MTR for network diagnosing.

[moltra]

I have started a couple of monitoring sites to help determine the cause of the reliability issues. I will update this thread with any problems that are found.

[moltra]

the more I look at the traceroutes, the more I think the problem is right at the node, or the switch that feeds the node.

Run 1
Traceroute #1 to 192.168.101.1 Hop Time (ms) IP Address FQDN
1 0.408 192.168.101.1 192.168.101.1
2 4.336 96.255.249.1 L300.WASHDC-VFTTP-85.verizon-gni.net
3 6.954 130.81.109.152 G0-3-5-7.WASHDC-LCR-21.verizon-gni.net
4 7.327 130.81.199.30 so-7-1-0-0.RES-BB-RTR1.verizon-gni.net
5 7.310 152.63.34.21 0.ae1.BR2.IAD8.ALTER.NET
6 7.236 4.68.62.137 ae17.edge1.washingtondc12.level3.net
7 17.236 4.69.158.22 vl-3502-ve-116.ebr1.Washington12.Level3.net
8 24.626 4.69.148.105 ae-6-6.ebr1.Atlanta2.Level3.net
9 29.946 4.69.137.149 ae-1-8.bar1.Orlando1.Level3.net
10 49.759 67.30.140.2 HostDime.com-10G-ethernet.core1.level3.net
11
12
13
14 37.130 64.37.56.252 64-37-56-252.static.dimenoc.com


Run 2
Traceroute #2 to 192.168.101.1 Hop Time (ms) IP Address FQDN
1 0.433 192.168.101.1 192.168.101.1
2 3.894 96.255.249.1 L300.WASHDC-VFTTP-85.verizon-gni.net
3 36.884 130.81.109.152 G0-3-5-7.WASHDC-LCR-21.verizon-gni.net
4 7.287 130.81.199.30 so-7-1-0-0.RES-BB-RTR1.verizon-gni.net
5 11.940 152.63.32.141 0.ae1.BR1.IAD8.ALTER.NET
6 9.846 4.68.62.133 ae16.edge1.washingtondc12.level3.net
7 9.369 4.69.158.22 vl-3502-ve-116.ebr1.Washington12.Level3.net
8 30.140 4.69.148.105 ae-6-6.ebr1.Atlanta2.Level3.net
9 29.517 4.69.137.149 ae-1-8.bar1.Orlando1.Level3.net
10 62.474 67.30.140.2 HostDime.com-10G-ethernet.core1.level3.net
11
12 36.916 64.37.56.252 64-37-56-252.static.dimenoc.com


Run 3
Traceroute #3 to 192.168.101.1 Hop Time (ms) IP Address FQDN
1 0.467 192.168.101.1 192.168.101.1
2 4.462 96.255.249.1 L300.WASHDC-VFTTP-85.verizon-gni.net
3 7.147 130.81.109.152 G0-3-5-7.WASHDC-LCR-21.verizon-gni.net
4 7.313 130.81.199.30 so-7-1-0-0.RES-BB-RTR1.verizon-gni.net
5 22.140 152.63.32.141 0.ae1.BR1.IAD8.ALTER.NET
6 9.788 4.68.62.133 ae16.edge1.washingtondc12.level3.net
7 14.759 4.69.158.22 vl-3502-ve-116.ebr1.Washington12.Level3.net
8 25.407 4.69.148.105 ae-6-6.ebr1.Atlanta2.Level3.net
9 31.571 4.69.137.149 ae-1-8.bar1.Orlando1.Level3.net
10 49.792 67.30.140.2 HostDime.com-10G-ethernet.core1.level3.net
11
12
13 34.224 64.37.56.252 64-37-56-252.static.dimenoc.com