New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
MyCustomHosting Down ? (Resolved. Issues? Open a ticket first!)
This discussion has been closed.
Comments
Trust customers to restart???? Read 90% of the threads on here and think about that.
For KVM I have no idea, it took me 3 minutes to sort the LittleHappyCloud KVM node which has around 50 VM's on it.
Xen was a little bit more complex because the CentOS4Xen maintainers are the worst in the known universe, i wont go in to to much detail but you are left in a position of reboot the node or clients wont be able to reboot at all.
I also did not receive the first email, but did receive the post-failure emails. And no, the first one wasn't spam-blocked, there's no record of it every hitting my mailserver.
It doesn't really matter, I don't fault Phil for his actions or subsequent handling of the snafu.
But RAID 0? I didn't know that until now. Very disappointing.
KVM on RAID0 is just... Terrible.
with a write back cache though, its about the least stable setup I can think of, I would be fine with it if I knew that is what I was buying in advance and it was sold as a pure performance product otherwise its just irresponsible.
Other than this disaster, my experience with MCH is still one of the best I've ever had on LET. RAID0+cache is obviously risky, but day to day performance is worth the risk IMO when you have proper backups.
I do agree that the product should be advertised accordingly tho. MCH stability and performance are very very good, but with the current setup, disaster recovery is very limited.
@agonyzt indeed I think there is a market for raw performance out there when redundancy is not important however if that is your business model every day you stay up is good luck not good management.
With this in mind the host should have a <1 hour complete recovery plan to get the node back up and all VPS recreated (not restored)
@AnthonySmith Indeed! That's why I would never host mission-critical stuff on this kind of servers. You don't run a business on a 24$/y VPS
Similar thread after heartbleed.
Well, what I had with MCH was certainly important (though perhaps not 'mission critical'). So I had it mirrored to a second VM (different provider, different location, different virtualization). When MCH went down I changed an A record and was back in 'business' in a few minutes.
Yes it costs me more that $24/year -- $36/year or (2 X $18) to be exact
My opinion about this situation:
Customers are responsible to the data inside VPSes, not providers, even if providers say "hey we always backup your data and will make sure there is no data loss", you're still responsive to your data.
Even if you use RAID 10, the system can fail, and all data can be lost.
Providers need to inform customers what is happening with their services, and what is happening with their data.
@comXyz I get what you're saying, but I wish I would have known that it was a RAID-0 setup before I purchased, because I never would have done so. I have proper backups, but to me the extra speed just isn't worth risk. RAID-1/10 isn't a backup, but it greatly reduces the risk of data loss.
I agree with that in principal yes, but I also think that providers have a responsibility to provide an infrastructure that can survive a simple unplanned reboot or panic, while raid 0 probably can manage this 9 times out of 10 when you add a write back cache on top, that number gets a lot less favorable.
^^ This
Automated or manually done?
maybe he can restore the data from the nightly backups. that way only 1 day worth of data is lost
I agree with you also, if the providers use RAID 0, they should have it bold in size 20. If ti's RAID 1 or RAID 10, they don't need to say anything.
If I know it's RAID 0, I will never sign up.
Manual. Though now I know how to automate it
You must have a short TTL for it to be an effective change within minutes. Apparently not all DNS servers would honour such a small TTL.
minutes? that long?
This^ +1 - its so cheap and easy to do with LET.
>
Does OVH BHS rent dedicated servers with more then 3 drive bays ?
Maybe nobody selling VPS in BVH have RAID10 ?
EDIT: I have always been aware that MCH was RAID0. I was not affected by the current issues, nor have I been affected negatively by the RAID0 setup during the almost two years that I have had service with MCH.
With their Infra range, you can order the specs you want.
EDIT: You need to call your OVH representative
got it few minutes again:
"_At this point we have all the VM's recreated. You may login to SolusVM and start reinstalling your VM's. The next goals will be bringing up the 2 in house DNS servers and the CA webserver. This is expected to be completed by the end of the day. The billing system and main site will be brought online after this is completed.
Again, sorry for the trouble with VM data. I do not provide customer backups at this time due to issues experienced prior with SolusVM's backup system.
The CA webserver lost no data. We provide clients with daily offsite backups with this service and data will be restored when the time shall come. Again, expected completion is end of day_"
Phil Robinson
My Custom Hosting`
Anyone have the link to login to the VM panel? I'm away from my regular PC and forget the address.
raid0isbae.mycustomhosting.net
https://github.com/blackdotsh/UptimeFlare
https://github.com/blackdotsh/StatusCake-CloudFlare
And your sig is
C'mon get sharing
Die, please die. I hate that word.
Sorry xD
Oh, my, this seems like an epic disaster. All the content of the three KVM nodes gone? Wow. :-(