Server Downtime: The Impact

The TweetLater website was unavailable for roughly 20.5 hours, from around 4:00 PM EST on April 16, 2009 to 12:30 PM EST on April 17, 2009.

Scheduled Tweets

All scheduled tweets that were supposed to have been published during the downtime have been placed into an error condition. They will not be automatically published. You need to review each one and decide if you want to reschedule it, or if you want to delete it.

Follower Processing

Follower processing has resumed where it stopped when the server went down. In other words, followers who have followed you during the downtime will be processed. It will take roughly 8 hours from 12:30 PM EST to cycle through all the Twitter accounts.

Profile Banners

We have already sent credits to the two people whose profile banners were affected by the downtime.

TweetLater Professional Subscriptions

All payment information sent by PayPal to our system during the downtime have been manually applied to the user accounts.

If any TweetLater Professional customer requires a partial refund for the downtime, we will be more than happy to oblige. On payments of $19.97 the credit is $0.57, and on payments of $29.97 the credit is $0.85. Please submit a support ticket and let us know if you want a credit.

TweetLater Professional Trial Members

We added two additional days to all TweetLater Professional trials that were active when the server went down.

Lessons Learned

  1. Communication under these circumstances is vital. We’re blown away by the understanding and support that folks expressed on the other blog post that gave status updates during the downtime.
  2. Hosting your blog and support system on a different server is very important, so that they are available even when your main site is down. This is something we put in place a little more than a week ago (thankfully).
  3. We’re just again reminded that we’re working with a machine that can break. These server disks normally have a life span of 50,000 power on hours. However, just like any other piece of machinery they can break.
  4. An attribute of crap is that it usually hits the fan at the most inconvenient times.
Advertisements

6 thoughts on “Server Downtime: The Impact

  1. Rebecca

    Congratulations and BRAVO and maintaining constant contact with your Tweeps…I'm sure I speak for all Tweetlater subscribers in saying – Thank you!

  2. Dewald Pretorius

    More than a month ago I submitted a ticket to iDevAffiliate, asking them to make it possible to host the banner graphics on a service like Amazon S3. Currently, the iDevAffiliate software forces you to host the banner graphics on the same machine and in the same folder structure as the software.I have had zero response to that support ticket.

  3. David

    Thanks for the contact throughout. Knowing that there was a light at the end of the tunnel (and that it wasn't a headlight) really helped!~

  4. 3CirclesChurchCommunities

    DP, Glad to see your support is top notch, it will make me feel secure when I start paying for the professional account in 6.5 days. As a fomer ISP who knows the value of good support… KUDOS all around!Lived in St. John's NB briefly back in 1974. Our doctor was a naturopathic, Dr. Bruce Hayhoe. Then moved to Bear River, NS and took the ferry over from Digby. Beautiful part of the planet.Leroy Grey, Nashville, TN

  5. Maurice Bizzarri

    Well I've been doing IT since before it was called IT. I probably have more experience in this stuff than you are old – 38 years. Stuff happens. Murphy was an optimist.Obvious (and probably unwanted) suggestion: RAID drives for your system on a SAN on a redundant system on a physically separate system. In other words, you should have 4 times the capacity you need divided into two separate places (physically). One stack of RAID drives failing should not bring down your system.Good luck. Thanks for such a good service.

Comments are closed.