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
New node will use old node HDD's ...
exquisite
You know it's premium when it's a 1G RJ45 feeding into each server, especially those used for storage nodes... that nodes gonna go brrrFFFFFFFFFFFFFFFrrrrrrrrrFFFFFFFFFFFFFF all the time
Plugged into daisy-chained 100Mbps mini-hubs.
https://www.aliexpress.com/item/1005004290105243.html?spm=a2g0o.productlist.0.0.5bae67790oIZyL&algo_pvid=dd28d20d-01c0-480c-904f-032e79791aa0&algo_exp_id=dd28d20d-01c0-480c-904f-032e79791aa0-0&pdp_ext_f={"sku_id":"12000028646102778"}&pdp_npi=2@dis!EUR!10.09!8.58!!!!!@2103399116606599178301436e43fa!12000028646102778!sea&curPageLogUid=f2ysBDL5KnmS
You joke but we know he can read this and think that it's a really good idea.
If this comment is not ok, I accept that the moderators delete / edit it:
A small update, because some people on this forum said that I plan to sell HDDs on olx and that I will disappear, which will not happen in any case
I worked from this morning on the 22nd until now on the 23rd at 1 AM , to put the new storage nodes into operation, and to test them, in the next 2 days all VPSs will be rebuilt, and 4 days will be given for the downtime
New node have this specification
OLD Configuration
2x e5-2450l 16c/32t
48 TB HDD SATA (raid 0)
192 GB Ram DDr3
NEW Configuration
2x e5-2670 v2 20c/40t
96 TB HDD SATA (raid 5) 84 TB free for usage
128 GB Ram (44 GB Ram free for ZFS Partition)
We have 3 nodes of this kind, and we managed to save 700 watts per hour with this new configuration
700w x 24H x 30 days500 KWh=aprox 300 eur save (electricity price aprox. 0.64 cents per KWh)
Regards,
Calin
lol
Are you running ZFS on top of raid5?
He’s going to really regret going to R5. It’s really not a good choice for VPS hosting.
Francisco
Raid5 is worse than Raid0 ...
Another obvious observation is that you have about half the RAM per core compared to what you had before. Maybe you just had lots of spare RAM before, but if not, this will seriously affect people's experiences of the new machines. I think people have joked before about you using swap and pretending it's RAM, but it really does kill performance because then your cores will be spending a lot of time being idle, and it will get exponentially worse as the load increases.
Sure more RAM uses more power, but it's probably one of the least significant factors compared to the CPU or disk.
Also, still not sure why you need 4 days downtime. You can migrate VPS one at a time, with just a small downtime for each one.
ZFS in hands of incompetent operator is ticking bomb. It is very specific FS and should be handled with care and deep knowledge of ZFS documentation. Hold on for that scrub and resilvering events...
It s hard for migrate lot of data , for example , now it s hard for rebuild VPS because i m receive lot of DDoS on backbone and gateway
Regards,
Calin
Left: @Peppery9 @Francisco @Hotmarer @ralf @LTniger
Right: @Calin
For write performance definitely, though there is at least some redundancy.
Possibly not enough IMO though: 86TB out of 96TB implies 8 12TB drives with a single parity volume (as is normal for RAID5) with zero hot spares (which are recommended for large arrays). I suppose it could be 16 6TB drives with one for parity and one reserved as a hot spare (if it were two for parity that would be RAID6 not RAID5), though I think that less likely.
Any power saved is likely to be eaten by the extra power consumed in IO operations when things need to be swapped in & out, would be my gut thought (caveat: I've not done any calculations to see if this is as likely to be true as I assume).
FTFY. RAID 5 Is a no go anywhere imo.
I hate to be the one beating a dead horse but I feel I must warn any other person out there with the slightest of hopes that IhostArt and @calin are not as bad as they seem, I recently got my fingers severally burnt by them.
I had purchased a VPS soon after he started his operations sometime in Jan 2022. the Vps in question was priced at 20 euro a year. Fast forward a few months and with a number of outages and 1 IP change I was planning to migrate off before the end of the plan, would of scheduled a migration plan for December 2022, still looking at a number of replacement hosts.
Suddenly on the 22 August the VPS was unreachable once again, gave it now attention as there is always a day or 2 down time every now and then. Today 25th of August I reach out to him to figure out why its still down, nothing related to my instance was in any Announcement of late, only to be met with news that my VPS is no longer available, the plan I had with them was no longer economically viable so they just shut it down with no notice and no time to backup anything at all. The only thing he can provide at this time is a refund for the remaining months.
Tried to reason with him to potentially buy another vps just to transfer out my data but no. absolutely terrible way of running a business. I know what you are thinking and you are right, silly of me to think I could be spared from such incompetence. Egg on my face now.
Buyer be ware.... nothing is safe.
Hello, on the 22nd of August we started to migrate all the storage vps, on some larger nodes, immediately on the 23rd-24th of August we were the target of a ddos attack, on the 25th our LIR with which we had IPs leased suspended my /24 subnet. At the moment we can no longer host clients per year for 1.6 euros per month. Since we have a very limited number of IPv4, we only have 96 IPv4 in our name, which are bought by us, we are trying to buy more IPs but they are very expensive, for example we pay 30 euros per IP
There were 3 incidents in one week, 2 unscheduled and 1 scheduled, the scheduled one is about VPS migration. As I said in the tickets to several people who encountered problems with the 20 euro/year VPS, we are willing to offer a refund for the remaining months, for some we have already offered them a refund. Those who have VPSs of at least 7 euros/month will be able to continue using their services and their IPs will be changed on our own subnet owned by us, not rented from any LIR
We have not yet completed the process of moving all the VPS storage to the new nodes, because we woke up overnight without any available IP, but we hope that in the next 2 days we will restore everything, and after that we will also talk with customers who had vps per year at the price of 20 euros/year for refund
Unfortunately, some unforeseen situations arose
We apologize for these delays, but for example, when we reach 70% of client migration, we noticed that suddenly we have to change the IPs of all our clients as an emergency, and at the moment we are overloaded with tickets and a lot of message
Regards,
Calin
But you removed those vpses without notice first just like that and when clients started asking about outage you responded them with refund for the remaining months because their vpses were removed already?
Wau, that's worse I imagined.
The data is present, we have a separate node for the vps per year, the problem is that we have nowhere to give them IPs to regain access to the VPS to make a backup
The only free IPs we have now are the IPs we have in our name, which are not rented, and they are in the process of being added to VPS storage
Regards,
Calin
There are several ways, even simple IPv6 tunnel (if you don't have native IPv6) would do it for most of your clients to reach their data. Not ideal but at least you wouldn't cut them completely.
What a shit show.
I mean we warned people.
We told them to stay away from Calin and his "hosting company".
People still bought it.
And then complained when shit doesn't work out in a professional manner.
Like... Make stupid decisions get stupid results?
If you're struggling to make the cash to pay for your IPs you could try to sell some of your wooden racks to an IT curiosities museum.
DMCA Ignored vps? That was easy.
https://netengi.com/vps
https://alexhost.com/vps/
https://www.yourserver.se/
https://ginernet.com/en/vps/
https://njal.la/servers/
the subnet of IPs was suspended for 400+ report in the last 10 month not because I didn't pay for the subnet. It wasn't about the money and why they had clients on those servers, but I learned from my mistakes, now I buy the IPs no renting for no problems in future
based and calinpilled
That pretty much sums up the entire company, yeah 😂
Just announce 192.168/16. Im sure your upstreams wont mind.