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
look like lv-shared03 gone bye bye
Sorry to hear about this, this reminds me of a post I made a while back and got banned when my temper flared at an admin. People think I'm crazy when I recommend or defend mirroring (RAID 1) as opposed to RAID 10. I'm not naive though, things can happen on RAID1 but so far, I've never had a problem knocks on wood. Plus it's just easier for me to restore a drive if something happens.
Wha?
No issues here on 03.Box is still up, looks like a Litespeed crash maybe?
Double check please IP pinging none of the site are working
how about lv-shared03 all my domains can not be accessed, httpd failed
I’m sure you created a ticket about it ?
The email from buyvm asked people to not open tickets unless they're still having problems after buyvm says the services are back up. Right now stuff is already known to be busted, so opening tickets doesn't convey any new info.
Common sense says that if there’s a new problem with a different server you should open a new ticket.
lv-shared03 is working fine for me as of now.
Yeah, sorry on lv-shared03. BuyCPanel had my licenses reversed (shared03's license on 04, 04's on 03) so when I asked them to release my license to reinstall 04...it reset 03's instead.
I was in the datacenter tonight working on storage as well as prepping for the future shared upgrades and couldn't hear my alerts going off.
Anyway, my fault on that. I worked with BuyCPanel to fix the discrepancy in the IP's and such so that headache won't happen again.
As of now we're at 90% restored. I'm syncing over the last batch of accounts and Anthony will get that rolling in the next little bit.
Overall around 24 hours to get everyone sorted. If we were using differential archives instead of the "rsnapshot hardlink folders" we could've shaved quite a few more hours off that i'm sure.
Francisco
@Francisco did you try that "one weird trick" that someone posted above?
We did!
It didn't work, at least not yet
We're going to try it again to see if we maybe need to do it longer or what have you. Going to try loading Samsungs Magician stuff and see if that can maybe fix it.
Francisco
Yeah, if this works I’ll be surprised much
We brought the drives back to the house so we'll run them for longer.
I'm not putting much hopes on it, i'll simply move forward with the changes I discussed earlier. It'll make everyone happy.
Francisco
2D TLC might have it's own problems (theoretical shorter lifespan compared to MLC) but 3D NAND, be it MLC, TLC or SLC will let you down if there is trouble with one of the die, even if most of the cells are good. While with 2D NAND you had some time to realize your drive was dying - if it was build around a good controller. 3D NAND is interesting: 32 layers on a chip is good for cost reduction, but that doesn't allow RAID 5 or RAID 6 style data allocation (chip level) as some controller can do. I guess they could still do that at the layer level. Not sure if it's done or not by most modern controllers, probably not (?).
If the controller burns, your screwed either way; could be what happened to you: would be pretty interesting to know what component went wrong in your SSDs!
Congrats for quickly fixing the problem though!
Was this on the buyshared or buyvm brand? I supose it was on buyshared and not buyvm, since buyvm does not offer backups.
If it was on the buyshared brand, they advertise on their website that all plans have "Nightly Backups". So how can the backup they have, be some days old?
Some backups could have failed, for example.
BuyVM in Vegas has some backups, we'll be offering full nightly backups on slices sometime in the summer.
The nightly was correct when we did r1soft but since that kept breaking we use JetBackup which can sometimes take > 1 day to generate backups making the backups not line up properly.
Sometimes Jet doesn't generate a backup for some users even though it should've as well.
Francisco
Ya my data restored was from Sunday
But hey, at least it was restored by them and saved me from having to deploy my backup which was also outdated.
Far as I know we've restored every account we have any backups of. Anthony will be auditing to make sure there's no backups from terminated users restored, but overall a job well done on his part.
Francisco
Backups fire on Monday, Wednesday, and Friday, but yeah, sometimes backups can drift for a long time.
The node we have storing everything is RAID60 and so the amount of IOPs it has available is pretty damn low.
Our plan right now is to do local RAID10's and return things back to Nightly again. I'll also use the stream/block level deduplication system we have for BuyVM to generate weekly off-node disaster recovery images.
As the month goes on we'll be scheduling an hour or two downtime on the other shared nodes in Vegas and rebuild them into exact matching configurations as well. Since I can just straight DD the data off the current arrays onto the NVME's it'll go quick.
Francisco
Check that your web server domain configs is up to date. I see your cpanel recognises domains that I had previous, but the front end web server isn't up to speed on them.
That might've been a configuration setting. By default cpanel doesn't allow you to have addon domains pointing outside of
public_html
but on the previous node we allowed that (that way you could have your addon domains in your root folder).As the imports ran it broke a lot of addon domains since it repointed them to
public_html
instead. You can just go toAddon Domains
to change the path's.Francisco
You really want someone else to look over your shoulder so you don't mess up target and source then.
Yes, apparently it was. Thanks for the hint.
@Francisco LET community will be waiting for you to publish a book titled "A little messier than I thought".
"Awww fuck."
"A guide to webhosting"
Francisco
Neat. Are these weekly backups made on LUX-shared nodes too? that would make me smile.
I'm just too lazy to backup friends n family shared hosting.
How is jetbackup configured if it can take more than a day to perform a backup of a server? o.O Even with a raid 6 array that I have for backups, 400 accounts take only a few hours and is mostly due to some accounts having 2+ million files :')
What? Mine takes a couple of hours to backup up around 15 accounts with 60GB in total. I have to check it.