SoylentNews is people
SoylentNews
[Update 5: All done. Nuttin but net. --TMB]
[Update 4: As of 20180509 @ 11:55 UTC, beryllium has been successfully rebooted. This leaves hydrogen to be rebooted in just over 13 hours. --martyb]
[Update 3: As of 20180509 @ 0414 UTC, both lithium and sodium appear to have successfully completed their reboots. That leaves beryllium (1hr45m from now) and hydrogen (20h45m from now) to complete their reboots. --martyb]
[Update 2: The second round of reboots went peachy keen as well. Next round starts at 3AM UTC (7 hours from this story's time) with our dev server (lithium). An hour later the load balancer (sodium) that I switched us off of this morning will reboot. Two hours after that the box (beryllium) that hosts the wiki, mail, IRC, and some other lesser-used stuff will get bounced. If you can't stand being disconnected from IRC for a few minutes, add irc2.sylnt.us (6667/6697) to the list of servers for this network. --TMB]
[Update 1: The first scheduled reboot (of fluorine) was successful. The two-hour reboot window for helium starts 1.5 hours from the date/time stamp for this story. Two hours after that marks the commencement of the two-hour reboot window for boron, magnesium, and neon. We do not anticipate any site interruption as a result of these reboots. --martyb]
We have been informed by Linode (on which all of the SoylentNews servers are hosted) that maintenance is required to mitigate against the Spectre (v1 and v2) attacks. As a result, all of our servers will require a reboot. Historically, any given server is down for anywhere from 15-30 minutes. We have redundancies in place for many of our operations, but there may be some unavoidable downtime. We ask your patience and understanding during this process.
The scheduled reboots are:
| Sat | 2018-05-05 1:00:00 AM UTC | fluorine [1] | Production Cluster | Completed |
| Tue | 2018-05-08 1:00:00 AM UTC | helium | Production Cluster | Completed |
| Tue | 2018-05-08 3:00:00 AM UTC | boron | Services Cluster | Completed |
| Tue | 2018-05-08 3:00:00 AM UTC | magnesium | Frontend Proxy | Completed |
| Tue | 2018-05-08 3:00:00 AM UTC | neon | Production Cluster | Completed |
| Wed | 2018-05-09 3:00:00 AM UTC | lithium | Development Cluster | Completed |
| Wed | 2018-05-09 4:00:00 AM UTC | sodium | Frontend Proxy | Completed |
| Wed | 2018-05-09 6:00:00 AM UTC | beryllium [2] | Services Cluster | Completed |
| Thu | 2018-05-10 1:00:00 AM UTC | hydrogen | Production Cluster | Completed |
[1] Unable to process subscriptions or update comment counts or deliver messages until it reboots.
[2] IRC (Internet Relay Chat) server will be unavailable.
(Score: 5, Informative) by The Mighty Buzzard on Thursday May 03 2018, @02:14PM (2 children)
For us? Servers. We've never exceeded our bandwidth cap that I'm aware of. I believe it's the sum of all of our allotted transfer allowances, regardless of which box they originate from, which should be 29TB/month by my math. All the internal IPv6 traffic between the db servers, web frontends, load balancers, and such costs us nothing.
We're currently running two load balancers at $10/month (though we could probably drop down to the $5/month package for them without issue), two web frontends at $40/month (whose specs might be overkill as well), two ndb cluster database node boxes at $40/month, one dev box at $20/month, one non-website outward facing services box at $20/month, one staff/inward facing services box at $20/month, and one backups box on another provider for I believe $10/month. And regular backups of the vms of our three most critical boxes by linode for $20/month total.
My rights don't end where your fear begins.
(Score: 2) by bzipitidoo on Thursday May 03 2018, @08:09PM (1 child)
So, $180 per month total? I surmise that such round numbers means those must be the prices of renting virtual computers from the cloud, or the cost for rack space for real hardware, and not the cost of electricity.
I've been looking at your Funding Goal, and wondering if there was more you could do to lower expenses. Last time I worked for a web services company, they had quite a few servers in a room in the main office. No reason to rent rack space for things that didn't need massive networking throughput, such as dev boxes.
(Score: 2) by The Mighty Buzzard on Friday May 04 2018, @12:55AM
Nope, $260 to Linode and another ten to our "offsite" backup server provider per month or a total of $3240 server costs per year. We've been covering that as well as the non-technical business expenses and managing to stuff a wee bit back for a rainy day*. The tl;dr of it is, I've no objection to seeing if we can trim things down a bit more but it's a big task and not a dire need right at the moment.
Keep in mind that we don't have an office we can just stick the less crucial servers in. They've either got to be VPSes, colocated, or sitting in someone's home office/bedroom with all the unreliability that implies.
* I don't know how much. Only NCommander and matt_ have access to the account. I think. Deucalion/Juggs may as well.
My rights don't end where your fear begins.