The TweetLater.com server is currently not responding.
We’re working on getting it back online as soon as possible.
If you happen to get to the login form, and it says your credentials are invalid, it’s because the database is down.
We will update this message when service has been restored.
Update 9:24 PM EST — The faulty drive is being replaced on the server. Unfortunately it was the primary drive, which holds the operating system. The OS needs to be reloaded, which they tell us will take between 6 and 8 hours.
Update 10:51 PM EST — The server rebuild is being started now. The database has been backed up, so there should not be any data loss. Follower processing and scheduled tweets are currently in mid-air suspension. When we reactivate the batch jobs once the server is rebuilt, follower processing will resume. All unpublished scheduled tweets and DMs that should have been published during the downtime will be published at that point. You may experience a sudden publishing of several tweets at the same time in your timeline if you had several tweets that should have been published during the downtime.
Update 11:04 PM EST –The hosting folks say that the server rebuild will take 6 to 8 hours. We’re going to assume a horizontal position for a few hours. When we’re awake and fresh, we’ll check that everything has been correctly configured on the server before we open TweetLater.com for logins. Will update here again in the morning.
Update 5:08 AM EST — The server has been rebuilt, and the backups are currently being restored.
Update 6:55 AM EST — Backups have been restored. There has been no data loss, apart from a few Keyword Alert archives that were corrupted beyond repair. It shouldn’t be an issue because those archives are only kept for 24 hours anyway. There are a few more configuration tasks left to do. It will probably be another hour, maybe two.
Update 7:02 AM EST — After doing an analysis of the scheduled tweets that should have been published during the downtime, we felt it would be prudent to put those tweets into an error condition so that they do not publish, and that you can decide whether you want to delete them or reschedule them. There are many Twitter accounts that will have a small avalanche of tweets published on them all at once if we just let the system publish the tweets. We’ll make a separate blog post regarding these tweets as soon as the system goes live.
Update 8:33 AM EST — It’s still going to be a while. Apache is still recompiling. Once that’s done, there are still a few configuration tasks to perform.
Update 9:14 AM EST — Apache recompile has finished. Now doing the last configuration tasks. It won’t be much longer.
Update 9:55 AM EST — Nearly there. Some of the batch jobs are already running, so you will see scheduled tweets being published. The site is still down because MEMCACHE is not yet running, and without it the site simply unceremonially crashes.
Update 11:01 AM EST — The hosting provider technicians forgot to compile several essential libraries into PHP. Just waiting for them to do that, then the site should come back online.
Update 12:01 PM EST — Just learned an important lesson. HostGator works on the tickets that have the oldest customer reply time first. When you keep submitting replies, such as “Please work on this, it is really important,” you bump your own ticket just further down the queue because the reply is new. So we learn a new thing every day.
Update 12:30 PM EST — And…….
WE ARE BACK!!!!