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.
How's Hetzner Cloud? Are random reboots common?
I have a few VMs running on Vultr, and they often reboot all the VMs. I run some services that are a pain in the ass to recover when the server reboots ungracefully. I'm trying out Hetzner Cloud and so far so good, I'd like to transfer my Vultr VMs to Hetzner but I'm concerned about the random reboots.
Vultr at least lets me know beforehand about the reboots. Does Hetzner do that? Do the reboots happen often?
Thanks
Comments
You should fix this. I'm sure you've heard it 1,000 times.
I can only recall a single instance where there was a reboot that wasn't initiated by me which was I think related to some problem with the host node it was on. (This was a cx11-ceph)
That's since about May.
@Corey nice idea in theory but some things can get really upset by sudden unexpected powerdown/reboot, particularly if a DB gets killed mid transaction it can sometimes cause problems.
With maintenance work You receive an email. I am a customer from the start and my instances are rebooted twice. You can see past maintenance and faults here: https://www.hetzner-status.de/en.html
Also with Hetzner expect routing issues one or two times a year.
My Cloud had no reboot since it's online (basically since it got released)
Never had a reboot and I'd be wary of a provider that randomly and without prior notice reboots your box(es).
Yeah, I know, but it's not my software and I have no control over it.
@thread thanks for the input, sounds like it should be solid, I'll migrate now. Thanks!
I've seen some reports of random Hetzner reboots though haven't experienced any myself. They will happen with any VPS. As someone mentioned, if your service is hard to restart then you should fix that.
I'm not sure if failover of a Hetzner Ceph VPS is experienced as a reboot to the user VM. Anyone know? Either way the container has to lose some context. If you need zero downtime then you have to do all the usual HA stuff with application level data replication, etc.
Then write cleanup scripts into your startup script of the application. There are tons of things that can be done from the 'outside' to make an application play more nice on hard power events.
You're just unlucky. I have zero reboots with Vultr in different US locations. But what about hezner, I see hell a lot of downtimes network and other things.
Hetzner cloud has been good for me although had a 1.5 hour outage a few days ago, when it returned the uptime had reset. Have asked for reason but got nothing back yet.
It was on their status page. Some power outage to some of their racks.
No, that's what the replay logs are for - if a DB gets killed mid transaction, it will just throw out that transaction during normal recovery.
Unless you surely disable all these safety measures that are built into various database software
Thank you
Depends on how damaged the underlying FS is.
Get many servers and spread your workload across them following cloud-native design principles. If losing a server/service hurts, there's room for improvement. You can architect things in such a way that you don't need to rely on individual components, it just takes some planning and a little budgeting.
To be fair this does require some flexibility in what you're running - either possibly using something else, drastically changing your deployment, or even rewriting some code (for in-house applications). The point is, you can cluster filesystems, databases, etc so that individual or small failures aren't such a big deal.
I am running highly unstable, still under development blockchain related code. In short I'm running nodes. The problems I have after a sudden reboot sometimes are related to the database, but most times they're related to something else. I appreciate the input but it's not something I can control at the moment.
You want crash-only design:
https://www.usenix.org/legacy/events/hotos03/tech/full_papers/candea/candea_html/index.html
Why don't try vm with dedi cpu rather than their cloud vm?
Vultr is fine as long you don't make constant high cpu load overtime,...
I actually do have a dedicated resources VM and it also gets rebooted:
Dear Customer,
Regarding the following subscriptions:
xxxxxxxxxxxxxxxxxxxxxx in New Jersey
We have detected a system issue with the node hosting the instances listed above. Our engineering team applied system updates and scheduled a brief maintenance window to perform a node restart.
Date and Time:
Sep-10-2018 07:00 UTC (Sep-10-2018 03:00 Local Time)
Please note: This event will reboot your affected instances. Your data and configurations will not be affected by the reboot.
This is the risk of using shared resources. It's inevitable. Dedicated server is maybe your only option. metadedi.net has a bunch of cheap dedis. kimsufi is my recommendation as I never got rebooted until now but there is no guarantee as well that in the future there will be no reboot eg. caused by power failure. Virmach uses rebootless kernel update so I think is good. I have a rock solid vps with virmach.
Maintenance reboots are only rarely ungraceful shutdowns unless your VM hangs when asked to shutdown.
I like/use both but Vultr is less likely to have the following that I've seen at Hetzner: random reboots, network outages, and once console just was black and nothing worked without restarting server. I've had this happen at least one on another provider with a very high uptime VPS. Vultr isn't perfect either, I do think over the years it takes the edge. Digital Ocean is good and unlike Vultr has bandwidth pooling. Ultimately if you want to avoid reboots a dedicated may be best option, Kimsufi are cheap.
I've been using Hetzner Cloud for about 7 months, only had 1 incident during that time.
Out of curiousity, why do Hetnzer VPS's lose their network connections or nodes? reboot so often anyways? I don't really use them for anything that needs reliable uptime but when it is online it's nice. They seem like a very small operation for the instability.
Last 3 months I had 1 downtime. I wouldn't call this often.
ive got a cloud on there..... stable asf ...
but will be closing it soon. cutting cost. lost my full time job .
I've been in the Helsinki data center since it opened. This weekend I've experienced some pretty shitty service quality. Hetzner cloud have halted/shutdown my server three times. Each time it has shown as Online/Powered-on in the Cloud Console even though it's been completely dead. No email notices or anything. Their online status page has listed outages for other Hetzner Online services but not Cloud. Other instances were fine, but this one instance really felt it this weekend.
After the third unexpected shutdown today at 00:05, I restarted the service and found that IPv6 connectivity has disappeared completely. Not only for the instance they've halted, but for all my instances. It has yet to recover more than 7 hours later. There still isn't any info on the status page, Twitter, and no info of any kind in the cloud Console. Hetzner appears to be unaware of the issues.
So to answer your question: no, they don't randomly reboot your server. They shut it down and don't let you know about it.
I'll document my experiences with Hetzner Cloud this weekend in greater details in the coming days.