VPS down since yesterday

[eUKhost.com]

RAID1, RAID5, RAID10 or any other solution would have never helped in this situation as this was due to bug with OpenVZ and same command was suppose to take full control of the CPU no matter what drive we would have used as main drive.

There was no other solution to this problem. This is a huge bug with OpenVZ and we are trying to find a solution for this problem to avoid this from repeating on any other node.

[DPS Computing]

Glad to see that eUKhost are working with OpenVZ to get the problem fixed so that it doesn't occur again.

Hope everyones VPS Hostings are now fully operational again .

[Fidget]

Well, I am glad I wasn't on that node. My VPS Hosting has suffered problems with cPanel's latest attempts to update their already hole ridden software (god knows who does their beta testing, but hell knows where the holes are for sure).

I'll be honest, I was openly critical of the short notice given before the recent server move (and the concept of physically unplugging a sever and driving down the road with it was unpalatable at the time), but the expertise is actually pretty good here. I've been with providers who are much worse on my road to greater internet prominence. However, I agree that proper communication must be employed to calm such short term untenable situations as this.

I'm pleasantly surprised to find out that VPS Hosting's on the same C class are stored on different drives though, not so many VPS Hosting providers would do that I may ask for another VPS Hosting on another node then.

Anyway, good luck with it all. Chin up, it could easily be much much worse. If you have mission ciritical website hostings, then you have to consider your setup - something to fall back on is a necessity in ensuring uninterrupted service on the internet. There is no such a thing as 100% in any single method on the web.

[pkearney]

Hi,

Mark, is there some sort of credit against our charges for a failure like this (the VPS Hosting pages say 99.9% uptime guarantee - for this month we are significantly below that).

Regards

Paul

[mephisto]

As compensation to the longer downtime than initially promised and failing to meet the 99.9% uptime guaranty I believe that all effected VPS Hosting users should be given one year on the base backup service.

http://www.eukhost.com/backup.php

I personally lost two contracts with new customers as they were asking me about what happens when the server goes down, and typically it does, but they felt that the period of time for it to come back up was not good enough and so went elsewhere.

[DPS Computing]

Are there actually any figures to show how much below the 99.9% uptime it is?

[ericsean]

well very approximately (in eukhost's favour), the downtime for me this month was over a day. So given that the uptime for June is 29 days out of a possible 30 days, the uptime works out at 29/30 * 100 = 96.7%

See it from another perspective, wouldn't you be disappointed if you bought a boat that guaranteed 99.95% "uptime", and found out that in reality, whilst swimming back to shore, that your actual uptime worked out to be around 96.5%?

I bought these services to have good connection, good uptime and be fault tolerant. As of this month, they are managing only 1 out of 3.

[pkearney]

Hi,

Well I lost 1 customer as a result of this. That costs me £150 a year.

My downtime was 30 hours, so that around 96%.

Paul

[mephisto]

For the current month, assuming that it stay's up for the remaining of the month, mine was down for 21.5 hours out of the 720 hours in June. So that's a 97.01% uptime.

[DPS Computing]

Thanks ericsean, pkearney and mephisto for those individual stats - it helps me to understand the situation more.

These figures are around 2.5-3.5% down on what they should be which I can understand that is a disappointment.

I have confidence in the fact that eUKhost will sort this out for you - hopefully Mark will be along soon to answer you questions and requests .

[eUKhost.com]

What problem you have faced recently was due to bug in OpenVZ which managed to make problem for the OS. Soft reboot or any sort of other commands could not be executed on the node due to this bug. We had no other alternative but to reload the node and restore backups.

You said that your losses have been over £100 but it costs only additional £3 per month to choose Virtuozzo.

[ericsean]

Originally Posted by eukhost.com

You said that your losses have been over £100 but it costs only additional £3 per month to choose Virtuozzo.

What difference does that make? You advertise as part of your services, a guarantee "99.5% uptime" - not "99.5% uptime if you pay £3 a month more for the more reliable Virtuozzo".

It's up to you to implement the technology that gives us a 99.5% guaranteed uptime. It is not the bug in OpenVZ that was the biggest problem regarding the downtime - it was the time taken to rebuild the node and restore the VPS Hostings, according to the several updates I remember pressing you for on the days of the downtime.

[DPS Computing]

It does seem like that this bug in OpenVZ has caused more downtime than there is usually. I assume that the downtime figures have been met for past months so maybe it averages out to be higher than this months total?

I can understand that reloading and restoring the backups can take a long time. I am sure that this is not a regular occurance and you shouldn't have these problems again. As Mark said he is working with the OpenVZ developers to fix the issues .

[eUKhost.com]

I have made necessary changes on http://www.eukhost.com/<a href="http...</a>_linux.php

It wont be possible for us to guarantee better uptime on OpenVZ nodes as the updates and patches received for OpenVZ are less in comparison with Virtuozzo. None of this Virtualization softwares are owned by eUKhost but swsoft developers are quite active in releasing patches and sorting technical problems with Virtuozzo. We cannot blame anyone at OpenVZ as they are offering OpenVZ for free.

Our licensing cost per VPS Hosting Virtuozzo license is £5 but we charge less amount to our customers as Virtuozzo brings more reliability, uptime and happy customers.

[pkearney]

Hi Mark,

Call me old fashioned, but is it even worth you offering the OpenVZ VPS Hosting if it is so problematic? I'm assuming there is a higher overhead for support over the Virtuozzo option. I just want a reliable solution. I took the VPS Hosting on your advice after the previous problems I was experiencing.

Up until now I knew nothing about Virtuozzo. If it is so much better than OpenVZ why haven't you sent a message to all OpenVZ VPS Hosting owners highlighting the unreliability of it and offering Virtuozzo.

You have amended the uptime guarantee for OpenVZ to 99%, but the last outage was less than that. So exactly what is the guarantee?

I'm sorry to be dragging this on, but I am unhappy enough that I am looking at alternative ISPs. I thought I had found a great ISP with eUKhosts, but the added functionality and options are made irrelevant if the service fails.


It seems things only really get fixed/answered when someone like yourself gets involved. I know you don't want to be on the frontline as much as your job now is marketing, but don't you realise that the fact that you are getting invloved means there is something wrong with the way it is being done now.

I think as a gesture of good will for a period of time (6 months - to equate to the costs of a months VPS Hosting) that you upgrade users that were on the failed node to Virtuozzo, after the 6 months they then get to choose if they want to keep ot or revert back to an OpenVZ VPS Hosting.

Regards

Paul

[eUKhost.com]

Everything was good with OpenVZ up till now but this problem which we have experienced on the node hosting your VPS Hosting was a huge bug with OpenVZ and none of us were aware of this problem.

As of now we have over 30 OpenVZ nodes so its going to be extremely difficult for us to get all customers moved to Virtuozzo. We have to find out some solution for existing OpenVZ nodes and so far we have succeeded in deploying some patches. We have added many restrictions on commands used by OpenVZ and tweaks have been performed on OS level.

I do agree that it was me who recommended VPS Hosting to you and that was based on my 1 year experience with OpenVZ. We had no Virtuozzo node at that time as pricing of Virtuozzo was stopping us from getting Virtuozzo in production and once we decided to go ahead with it then the whole procedure got delayed due to documentation work needed by swsoft.

We have team of highly talented people in our VPS Hosting Support team and one can definitely not challenge knowledge of our guys but no system admin in the world can do anything on a server that runs out of CPU resources due to one single command. There was nothing much that we would do at that moment and you can consider yourself unlucky as none of the other VPS Hosting nodes had such problems and it was only your node that had huge impact of this bug.

Reload process got delayed as lost couple of hours in attempt to resolve this problem and when we decided to go ahead with reload then it was previous RAID1 configuration which needed to be removed to have only 1 drive for restoration of backup on new drive.

Our guys did their best in that situation but the restore process could have never completed in lesser time. You are upset as you had to go through such major outage for more than 24 hours but you should consider yourself unlucky and nothing else as no one else had any such problem on any of the other OpenVZ nodes.

[mephisto]

While things about what's advertised and what's achieved, can anyone explain the differences between the CPU that is advertised "Quad Core Xeon x3210" and what's in my /proc/cpuinfo?

root@VPS Hosting202 ~ # cat /proc/cpuinfo 
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 15
model name      : Intel(R) Xeon(R) CPU            3040  @ 1.86GHz
stepping        : 6
cpu MHz         : 1862.011
cache size      : 2048 KB
physical id     : 0
siblings        : 2
core id         : 0
cpu cores       : 2
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 10
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc pni monitor ds_cpl est tm2 xtpr
bogomips        : 3726.40

processor       : 1
vendor_id       : GenuineIntel
cpu family      : 6
model           : 15
model name      : Intel(R) Xeon(R) CPU            3040  @ 1.86GHz
stepping        : 6
cpu MH