OVH web hosting - two weeks review




Our tale of woe and drama commenced on the morning of Thursday the 5th February when we placed our order for a Superplan 08 + dedicated server. We were so naive, we thought the hardware spec, the bundled support services (vKVM, free technician time etc) and the much hyped ‘IP Failover’ mechanism were terrific - along with a quick setup time (1 hour) and IP geolocalisation. “Give us one of those!” we exalted. Sure enough, OVH processed our payment lickety quick and within a matter of hours we received our root login details from OVH support. It started out so well, little did we know that hell in all its firey glory awaited with baited breath.

Our plans for this particular server involved having several virtual machines (VMware) running on the server, performing different tasks, one was to be a webserver, the other a mailserver while the host machine was to be a DNS nameserver. We immediately set about installing VMware Server, however we came across a rather ugly error. VMware refused to compile because the version of gcc used to compile the running kernel was version 3.* while the version installed on the server was v4.4.*. We’d never come across this, in all our RHCE’ing days, VMware actually managed to compile pretty much ok on all servers / hardware we’d ever used. So we checked the kernel and found we were running a ‘made-in-OVH’ kernel, basically a custom built kernel compiled by OVH themselves, this must be the much-famed ‘NetBoot’ system. We were under the impression that the ‘Net-Boot’ system was a rescue system which allowed the server to be booted into a functional kernel (albeit an OVH kernel) without the need to go directly into a rescue kernel.

So we simply went about switching the ‘Net-Boot’ setting to hd, so that the server would boot from the MBR of the server, and not be overridden by OVH’s kernels. This ought to be what we’d need in order to get into the official CentOS-supplied de-facto kernel. Oh how wrong we were. You see, OVH, in their infinite wisdom decided to bundle the CentOS installation with the LILO bootloader, while unusual, we had dealt with LILO before and proceeded to configure it to load the CentOS kernel which was downloaded using yum. Using the vKVM system we were able to monitor the boot process and the system managed to get through stages 1 and 2 of the boot process but failed miserably while trying to mount the partition mount-points in the third stage of the bootloader. As a result, the vanilla CentOS kernel just wouldn’t load. Let us not understate the attempts we made to try and get the kernel past the third stage. We went so far as to edit the init script within the initrd image, we tried pre-compiling and patching the hard-disk modules directly into the kernel - hell we even downloaded the kernel source and tried building our own kernel - all of which just resulted in epic fail. Just to be sure, we did all of the above again, only this time by using GRUB as our bootloader. No banana.

All the while we were doing the above, in between re-images etc, we fired off a few emails to the OVH support… All of the emails went unanswered. With OVH being a French company, one may be able to excuse them with a few indiscretion when it comes to translation. However, the word ‘Support’ is really such a simple word to comprehend, it’s criminal that they are allowed to use it to describe their service. In between our kernel panic affray we needed to get our additional IP addresses (OVH like to call these “Fail over IP’s”) online and aliased to the server’s ethernet device so that we could get our nameserver online by Friday evening (our particular deadline). One of the reasons we chose OVH was because their 1 hour setup time would’ve left us with plenty of time to get the nameserver online. So we logged into our (ridiculously badly designed) control panel to assign our so-called failover IP’s to our server, alas (and by now, unsurprisingly) there was a fault with the control panel which wouldn’t let us get any additional IP’s. We had one static IP, which was permanently bound to the server ethernet interface, and we were due 4 extra IP’s to use. We got on the phone to support about this one, since this was a critical problem, we had a deadline to meet, and while important, the kernel problem could take a back-seat to this.

We were told by the OVH support on the phone that we would need to request the Failover IP’s and that they would have to be added manually to the control panel/server, the reason for this apparently was that OVH’s systems had a fault and the work-around involved manual intervention/assignment. We were promised that we would have the IP’s within 24-48 hours - we’ve *never* had to wait more than 8 hours for an additional IP to be assigned before, but we thought we’d give OVH a little slack, since we thought it’d be likely that we’d have the IP’s in time to meet our deadline. The Thursday night came, and went, with much attempted kernel hacking in between.

Friday afternoon came around quickly, and we still hadn’t received our IP addresses, a horrible sinking feeling set in and we realised that we’d made a mistake in choosing OVH, a very big mistake. We got on the phone to support again and spoke to ‘Marco’ a French guy who told us exactly what he’d told us 24 hours previously, and that the IP’s would have to be assigned manu… blah blahfrenchaccentblah. We told him that under no uncertain terms that we needed the IP’s on that day or we’d be forced to cancel the server, since it wouldn’t meet with our deadline. Obviously something was escalated and we had a phone call five minutes later from an English guy asking why we wanted to cancel the server. I told him that we really didn’t want to, but the lack of support and IP assignment would force our hand. I told him that Saturday evening would be the *very* latest that we could extend our deadline to. The call ended and then he called back, about 15 minutes later telling us that it wouldn’t be until Monday that we’d get our “failover IP’s”. That would be a full 96 hours after the server came online - double that of what we were quoted as a *maximum* time. At that stage the cancellation was in the bag as far as we were concerned. We took out a dedicated server with our second choice of provider and were online (with our required IP’s) in time to meet our deadline. Thankfully there are dedicated server providers out there who take pride in the service that they offer.

We pulled an all-nighter getting the nameserver online (on the functioning providers box). We spent a few hours over the weekend trying to get the kernel to load on the OVH box, and trying to get OVH to respond to requests for help and information regarding the hardware in the server etc. OVH were absolutely useless, and so is every aspect of the service they provide. Even the ‘vKVM’ isn’t a true KVM since it won’t allow you to boot into the machine properly as it runs in a ‘virtualised’ mode or some crap. The weekend passed too quickly and we were now faced with the unpleasant task of getting our money back for the OVH server which was utterly useless and compounded by the failure to provide even the most basic of services, the additional IP’s within an acceptable time-frame. We spoke to Marco again, Marco told us that, in order for us to cancel the server and get a refund, we would have to write them a letter stating that we wished to cancel and post (yes, snail-mail) it to their UK offices in order for them to process the cancellation. I asked him if it would be possible for us to email over, or even fax a letter to them. Marco refused, simply stating that it was in their Terms and Conditions. I asked him what it was that specifically required us to send them a posted letter to cancel, he said that it was so that they knew it was me, the account holder who wished to cancel the contract. I pointed out the ridiculous nature of that assertion, since anyone could write a letter and they wouldn’t have a clue whether it came from me or some random person. Regardless, we sent out a letter, recorded delivery, first class mail, which would’ve arrived at their offices on Wednesday 13th February, 6 days after the server was paid for. We waited for OVH to process the refund, however, every day we checked, and we could see that the dedicated server was still available for us to configure etc (*still* without IP’s I might add), we still had root access to the box. That sinking feeling sank further. It became clear as day to us that OVH were not a reputable, honourable company in the least. This one was not going to be resolved easily and amicably as it could, and should have been. Nearly a week passed since OVH received our letter, more than ample time for a card refund to go back into our account, and still, nothing.

We gave their 020 support number a call again, and spoke to Marco who informed us that the person that deals with refunds and cancellations was ‘unavailable’ and told us to call back an hour later. We called back, and Marco told me that as far as they were concerned, they had processed the refund. This struck me as being a little odd, I asked Marco why I could still access the server via SSH and why it was still available in my control panel for configuring - Marco came up with a suitably rubbish explanation. I asked which date the refund was started on, he told me Wednesday the 13th Feb saying that it would take three days for it to come back into our account. Let’s count those days… Wednesday == one, Thursday == two, Friday == three, Monday == four, Tuesday (today) == five. That’s two days overdue. We contacted our bank, they have nothing on their radar for anything due to hit our account, card refunds or otherwise. Marco said that he would email over a receipt of the refund being processed by OVH’s bank, I asked him to do it while I was on the phone so I could view it there and then, Marco agreed and put me on hold, after three minutes he cut me off. Miraculously, the refund ‘receipt’ never arrived, (which is against English consumer laws I might add). I rang back and spoke to an English bloke this time, i went through the whole situation again with him, and he assured me that Marco would send over the receipt immediately, I told him that if I hadn’t received the receipt within an hour of that call, that I’d call back and ask them to send it again. It never arrived so this time I recorded the call, I really have to say that Marco is quite rude, he really needs to learn that it’s very rude to interrupt when someone is talking - I don’t know if that’s just his personality or whether people in France are generally just that rude. Anyway, back to the point. Marco said that he asked the billing department to get us a receipt, he said he needs to send it over as a link in an email to a screenshot of their billing system, this time, we have no refund, no *receipt* of a refund, and no idea when we can expect our money back.

The moral of this story, is always do extensive research about a company before you part with any money, read reviews, and never *ever* get a dedicated server from OVH unless you like headaches.




Leave a Reply

You must be logged in to post a comment.