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.
Comments
SBG1 was also affected by the fire: four of its rooms were severely affected (racks 61Bxx, 61Cxx, 62Bxx and 62Cxx)
If your server was in one of these rooms, its dust.
https://help.ovhcloud.com/en/faq/strasbourg-incident/which-datacentres-are-affected-strasbourg/
Yep, though it isn't just about complete duplication of resources, unless your time-to-recovery needs to be near zero.
It is rare that I don't RAID 1 anything on a server these days.
But of course RAID is not more of an availability solution than a backup solution, as this event proves there are things that will take out all the local drives in one go.
It doesn't have to be a full server though - just a storage account somewhere from which you can restore data from to a replacement server. Assuming of course the time-to-recovery of that is acceptable.
Again, your storage account is easy to have elsewhere. Though this slows recovery (internal bandwidth & latency in the same DC will likely be better). Heck for individual users protecting their own data rather than larger scale enterprises the off-server backups could be at home.
If the world is gone, you have bigger problems than where your data has ended up!
Unless you were promised a particular DC, a refund might not be an available option.
If you can't choose and aren't happy with all the possible options, then you either need to pick a different country or a different provider.
Exactly. With so many options available for spreading your backups for greater reliance, why keep them all near each other. Assuming of course you have multiple backups. Or any, which was not the case for some people affected by The Event.
https://www.ovh.com/fr/images/sbg/Octave-Klaba-speaking-en.mp4
Good, fucking, job.
--
Oh. Firefighters had termovision/FLIR, the UPS7 & UPS8 was on fire. UPS7 had maintenance work in the morning and
supplier
changed a lot of parts.Show that to people who are asking for data from SGB2 back on Twatter.
No, I did not claim that. I deduced that they had no adequate and effective fire-extinction system. And that is what the events there seem to confirm. But feel free to show me evidence that they did have an adequate and effective anti-fire system and let their DC unit burn down anyway, for whatever reasons.
I need to show evidence? No I do not. There is ample evidence to see for everyone.
Is this somehow a "religious" issue for you? Yes, some things are indeed very expensive in France (e.g. food) ... but running a colo is not one of those.
I might be wrong as there might be places where they charge for air, but in Europe air is free for everyone. And the cooling system you mean still is irrelevant in this case because those systems (basically flaps and ventilators) can close the intake e.g. if there is a fire.
That's how you transfer everything to the cloud.
I hate when people complain for their $3 vps for becoming french toast. You didn't have backups? Move on, why create more drama? If datacenters backed up your data to 10 different locations it wouldn't cost fucking $3. It would cost much more. Even OVH shouldn't give free months to those, they should only give what their customers paid for.
Technically, OVH is free from any SLA compensation. Their ToS states that fire damage is not covered by SLA.
But pretty sure OVH will give some free months as a mean to control damage.
My guess is once the servers are up that they have pending (oles said 10,000 servers), I presume they will triage based on most valuable to least valuable customers (over some time interval). Seems you'd want to preserve as much business as possible.
They already announced they would give some in coming weeks.
I know they said they would. But they are not entitled to. They are pretty safe from any potential lawsuits.
But what I understand so far from this incident are:
1) People won't create backups and cry later
2) They don't have clear knowledge on what a disaster recovery plan is
3) They don't have clear understanding on cloud(Physical premises part)
4) If Let servers caught fire we don't have to worry at all, we have LowEndFiremen
5) The end is nigh
90% people never backup. That is just a fact of humanity.
Can't blame them. This sort of stuff don't happen frequently.
That's what happens when you don't get out much and lack common sense.
Attaboy. You are on the right path.
“There is great value in disaster. All our mistakes are burned up. Thank God we can start a new.” Thomas A. Edison.
What backwards country doesn't have fire suppression requirements by law? That's some 1970's thinking.
It's cheaper.
that sucks.
So all the lowend politicians and newly minted fire suppression experts have something real to read rather than making up random arguments:
https://cfpa-e.eu/national-regulations/regulations-france/
https://cfpa-e.eu/wp-content/plugins/pdfjs-viewer-shortcode/pdfjs/web/viewer.php?file=https://cfpa-e.eu/wp-content/uploads/2019/05/CFPA_E_Guideline_No_14_2019-F.pdf&download=false&print=false&openfile=false
I know I know, googling gets in the way of spurious arguments - but oh well
The law will demand that the workers nearby others are sufficiently protected, and they were: no one was hurt and nothing beyond SBG* was on fire. Fire suppression is a detail that the legislation won't necessarily demand if other sufficient preparations are in place.
It is different for residential dwellings and public places, though even then things are a lot more iffy than you might like (flammable cladding on the outside of blocks of flats being a-OK until recently in the UK for instance).
That's mostly about living quarters and buildings with lots of people inside (e.g. shopping mall).
Whatever the legal requirements happen to be they should be regarded as a required minimum wrt DCs which have their own standards on top (e.g. Tier X). That said a DC can of course fall back to the legal minimum, but then they'd better not talk about Tier 3 and higher - which OVH to the best of my knowledge didn't do; they, so it seems, care little about DC standards.
2nd link was specific to DC's but oh well. There is also links at the site for UPS etc. I am sure the fire inspector report and insurance litigation will provide more info than everyone on the internet with zero access to the site
Thanks but those are just guidelines and recommendations. They are binding only if demanded by an insurance - which may or may not be the case wrt OVH; as you correctly state we do not know that.
In a video today, OVH's CEO said that after reviewing camera footage from the affected area of the datacenter, the origin of the fire was in two UPS (uninterruptible power supply) units. He added that these had just been serviced, possibly I imagine because of some concern about their operation. (And now I can't find the link again for the video. Maybe someone knows where it is.)
But as an old fogey, I remember factory telephone/computer system battery rooms that were filled with big banks of heavy lead-acid batteries that had to be load-tested periodically. Those as far as I can remember were not considered a fire hazard. Neither are banks of NiMH batteries, like in some hybrid cars. LiFePO4 chemistry presents some possible fire danger, but the big risk is in using high-capacity high-rate lithium-ion batteries that have a lot of cobalt in them to maximize their capacity. It would be interesting to know what kind of UPS batteries these were.
@bobbr
https://www.ovh.com/fr/images/sbg/index-en.html
It would have been lead-acid batteries. There is no way OVH would afford lithium (let alone LiFePO4) UPS. Those are much more expensive.
Oles looks like he hasn’t slept since
When receiving millions in investment for hosting business, companies should primarily use it to secure their datacenters, servicing fire-alarms, direct fire suppression system, having fire-resistant walls, and little tit-bits that help to protect, guard and shield the servers and millions of dollars worth customers' data (BTW the irony is: with such low end service provider will any one host their most critical data? Probably no! But data of all customers together can be treated as quite a worth!) that resides there than spending it on marketing overly-hyped "cloud services". It takes years or decade to build a reputation which can get destroyed by a nightly catastrophe!
I thought all datacenter rooms must have fire protection system.