Frequently Asked Questions and Answers
How to determine the initial number of Cores I would need?
As you may know already, we don’t limit the number of sites you can host per Aegir instance and it can automatically scale up to 32 GB RAM and 24 CPU Real Threads for short load peaks – free of charge. Every Aegir SSD+SAS Core, now available in Santa Clara, Dallas, New York City, Amsterdam, London and Singapore comes with 512 MB of database disk space, 10 GB of disk space for files and 300 GB of web in/out traffic generated monthly. You need enough Cores to accommodate expected usage of those resources. Since the traffic generated is most often the least important part of resources included per Core (most sites never use that much), unless the site is expected to serve some really big traffic (which happens very rarely), it should be enough to simply check the values of those two other limited resources: database binary space and disk space for files. While checking the disk space used by all files belonging to the site is rather simple (just check the size of your site files directory), determining the binary size of the database is a bit less easy/obvious, because it refers to the physical size of all binary files used by the database, which can be checked with command like du -s -h /var/lib/mysql/dbname, however if the database uses InnoDB tables, this check will be reliable only when there is non-default mysql config option enabled: innodb_file_per_table=1, as otherwise the real binary size of the database is hidden inside the globally shared /var/lib/mysql/ibdata* file. It is important to know that the database binary size is unrelated to the database size reported by mysql itself. It is also unrelated to the size of the exported database dump file. Note that you could drastically lower your database binary space usage by converting your tables from InnoDB to MyISAM. We have seen database binary space lowered to 1/2 or even 1/3 of the original size after converting to MyISAM. Your results may vary. Using InnoDB is important only for tables with high write-concurrency, and because we disable dblog/watchdog and “outsource” all cache* tables to Redis (cache_form bin is the only exception) by default, it is no longer recommended nor required to always use InnoDB. This trick alone can give you 2x (or more) database space allowed per Aegir SSD+SAS Core, effectively. If you are still unsure what are your needs, simply start with one Core, import your sites and wait until our running daily system monitor will notify you that you are using more resources than allocated and you should upgrade. You can always easily check your current usage, because it is displayed in the footer of your Aegir control panel.
How fast does your support team respond to incidents? Do you have a system in place to monitor website availability that will alert you if a site is not responding?
It depends on what you mean by “incidents”, but if this is anything on our side affecting more than one Aegir instance and not related to your sites internal issues, we generally respond even before you realize there is an issue, by posting status updates on Twitter and App.net. Also, as listed on our website we have both local, autonomous auto-healing systems and remote monitoring with Scout and Pingdom. We monitor remotely Aegir control panels availability only (every minute), but it translates to all your sites availability, because the control panel runs on the same Aegir instance/system.
Will you get notified if MySQL or Nginx crashes due to a traffic spike and restart them proactively?
It never really happens in our hosting, unless there is scheduled maintenance or some really serious problem like hurricane impact, but thanks to our built-in autonomous monitoring and auto-healing all services are monitored every 10 seconds and restarted automatically, if required for any reason.
If there’s an issue with the server, what contact methods can I use to reach support?
All support requests must be submitted via our website at http://omega8.cc/support because it allows us to log them properly in all our three support backends we use to guarantee reliable support. You can then follow up either via e-mail or via the helpdesk at https://aegir.happyfox.com. If you can’t access our website for any reason, like you are away from the desktop computer, you can also send the request directly to support AT omega8.cc and if the problem is really critical, like confirmed system down (and not just WSOD on your site because of error in your code, or your IP blocked on our firewall due to port scan or load testing, which are not allowed for security and systems stability reasons), you can also ping our NOC directly via our TEXT/SMS gateway nr we can provide after you sign up, for immediate assistance.
What if that box goes down or encounters an unrecoverable error? Does the box spin up on a new host box, and keep functioning?
Since we are using local virtualization on hardware we control in 100%, but without remote filesystems, in case of hardware failure it will not just fire up your instance on another hardware automatically, as it is possible with SAN based or any other HA setup. It would require using one of our duplicate R1Soft backups to re-create your system as-is on another running machine in the same location and start it from there. While this may sound not as cool as SAN based “cloud”, in fact it is much more reliable, because hardware issue will not affect other machines, while any serious issue with SAN would take down all connected machines, and this is what in fact happens at all known cloud providers. We are using hardware configuration which is much more reliable and allows us both to guarantee real separation and faster recovery when anything like that would happen. I would also recommend to check details in some context, as explained in those two articles: http://omega8.cc/compare/speed and http://omega8.cc/compare/scalability
My another concern would be overnight, if the box died and needed to be rebooted or replaced. Would incidents such as these be handled by your emergency support team?
As you already know, we monitor all systems 24/7 and we are ready to work on fixing any possible issues immediately. If there is an issue beyond our control, like hardware (network or server) problem and we can’t re-create your system on another working machine in the same location for any reason, our datacenter provider guarantees, using their own words: “that all production hardware, including servers, switches, load balancers, firewalls, and power distribution units, will be free of hardware problems. If a hardware device suffers full or partial hardware-level failure that causes the device to stop operating in an optimal manner, Voxel will replace the hardware within one (1) hour after the hardware has been determined to be at fault.”
Who is your data center, infrastructure, hardware and network provider?
After 15 years of experience with most of big names in the professional level hosting industry our choice is Voxel, just recently merged with Internap – they are in business only one year longer than we are, and we have a solid, good relationship with their top level team and all great sysadmins and engineers. We highly recommend them to anyone looking for a top quality, reliable provider.
What happened to your previous Mini, Classic and Standard options?
The old Mini has been renamed to one Aegir Standard Core, while the Classic has been renamed to four Aegir Standard Cores. They offer still the same (or better) performance with some greatly increased limits for allowed disk space, memory and CPU power. Also the price didn’t change for the Mini equivalent (1 Standard Core), while the price for Classic equivalent (4 Standard Cores) was a bit higher, but it didn’t affect any existing subscriptions, only new orders. As of August 11, 2012, Aegir Standard Cores have been replaced with Aegir SSD+SAS Cores, and this upgrade also didn’t change any existing subscriptions, as announced on May 5, 2012. Effectively, even if all existing Mini, Classic and Standard subscriptions for Aegir instances hosted in San Jose, Santa Clara, New York City, Amsterdam and Singapore are already powered by the same SSD+SAS technology, their billing for renewals didn’t change – they are simply no longer available for new orders. As for allowed database disk space, disk space for files, R1Soft/Idera backup space and web in/out traffic generated monthly (and the subscription price), all existing subscriptions for discountinued non-SSD Cores are effectively 1/2 (half) of Aegir SSD+SAS Cores.
What happened to your London location?
As of April 14th, 2013, we offer the same SSD+SAS technology in London UK, so we have added London again to the list of our six Aegir locations worldwide.
What happened to your Aegir Root Cores?
Aegir Root Cores are powered by our previous Cloud/SAN/iSCSI technology and have been discountinued. While they are not available for new orders, we still support all existing Aegir Root instances.
Is the Remote Install and Support still available?
Yes. We have moved it to its own page. You can find it here.
Do you really host only Drupal sites?
Yes. We specialize in the high performance and Drupal only hosting.
Do you offer also E-mail and DNS service?
No. We focus in 100% on Drupal web-hosting service in the SaaS and PaaS mode. We recommend to use DNS management tools provided by your registrar and Gandi Mail or Google Apps for your e-mail. For your Drupal sites outgoing e-mail we provide standard sendmail/Postfix local server, but we highly recommend to use your own SMTP server (provided by Gandi Mail or Google Apps) for better delivery rate, because some e-mail services like AOL don’t accept mailings sent from the site / web server directly, while your domain DNS MX records point elsewhere.
My site layout is broken because it still references already purged, aggregated css/js files!
You may want to use the AdvAgg module to avoid this. See also this hint.
How can I purge the Speed Booster cache?
It is no longer needed after we have switched default Speed Booster cache TTL to just 10 seconds (and 1 hour for known robots).
Is it possible to use Drush 5 with BOA now?
It is not possible yet, because the Aegir 2.x fork we are using in BOA doesn’t support Drush 5, because we would have to drop Drupal 5 support first. We plan to implement Drush 5 support as an option soon.
Where can I check servers or network status?
We require and expect excellence, so we accept only one possible status: all services are available without issues. If there is some unexpected issue affecting more than one Aegir instance in any location, we always post live updates on Twitter. Also when there is any scheduled maintenance or system upgrade, we announce it on Twitter.
Yes, we offer limited shell in all our options since October 2010. It allows you to run almost all possible Drush commands plus useful set of standard shell commands. It supports also rsync, scp, git, svn, cvs and many others.
I need some help with my current site but I’m not ready to purchase hosted Aegir, can you help me?
Yes, you can purchase our billed hourly basic help and later decide if you prefer to use our hosted service or consider it in the future. This basic help doesn’t include troubleshooting, only basic consultation and reviewing your current setup and possible solutions.
I don’t know which registrar offers advanced DNS management, can you recommend one?
Yes, we recommend Gandi.net. They have everything you need when you are using our application hosting: full DNS management plus free and big 5 E-mail mailboxes.
What is your Support availability and SLA guarantee?
We are working hard to keep your real uptime near 100%. Every Aegir hosted instance comes with our default support available Mon-Fri 8AM-8PM UTC and emergency support 24/7/365.
What is your cancellation and refund policy?
All setup and one-time fees are non-refundable. Also monthly fees, starting with second month are non-refundable. If you will pay for 3 or more months in advance and you will decide to cancel your subscription, and the service is available also in monthly subscription, you can do that at any time and all your unused balance will be returned to you, excluding current month. Note however, that the unused balance will be calculated using the price of the monthly subscription for the same service and not the price of your subscription, which typically includes substantial bonus for the pre-payment, but if you will cancel your service (with any subscription length) in its first 30 days, you will receive a full refund. For more details please read also our Terms of Service