Having learned a painful lesson from losing my entire database(s) when 123-reg so rudely deleted my VPS, I have implemented a simple system for backing up all my live data from the new VPS.
I have set up cron jobs to run scripts on the remote VPS to back up each database at a specific time each day, and I have complimented this with a cron job/script on my home server to pull that data off at a specific time, daily.
It’s probably not the most robust backup system in existence but it does the job, and it only took me a few hours to sort out. It’s infinitely better than no backups at all anyway.
Now I will soon be free to crack on with MakingTracksGPS…
So, I went ahead and released GeeOhTileServer v1.0 a few moments ago. We. Are. Live.
Check it out here!
I thought it was going to take a while longer than it did, so I initially pencilled in the 13th May for the release date, but here we are – ahead of schedule.
Now everything will be geared towards integrating with MakingTracksGPS, and making some strides towards the fuller implementation of Gee-Oh!
So, I managed to survive the 123-reg disaster, and I’ve migrated my VPS across to 1&1, sans MySQL data, and basically gone with a fresh install. I got an email, after a week of “we’re doing all we can…” emails, to say “soz, we’ve deleted it all and it’s gone” followed up with an offer of 6 months free hosting and a year’s free backups. No tar.
I’ve still got other services with them, but the VPN is too important. The new one is a bit more powerful anyway, which will be useful for when I put GeeOhTileServer live (very soon hopefully).
Still got a few little tweaks to make, but essentially I’m all up and running again. Boss.
I just mean the coastal matching of the base land map, not that the country as a whole is fucked – although you could argue that might be true also…
Anyway, the auto-processing of entire GridSquares has been such an improvement, cutting my processing time down by a factor of 20 or 30. Very pleased with that optimisation. I’m happy to release this map as part of the first version of GeeOhTileServer, on the understanding that only the UK/Eire has an accurate land mass.
Since my VPS on 123-reg has now died, I obviously can’t proceed to release it live (for now), but I will release it on my home server in the meantime, and possibly use that to develop MakingTracksGPS against.
I have yet to decide what I’m doing about 123-reg, but in all likelihood I’ll be migrating my VPS service away from them. I’ll be keeping them in terms of handling my DNS for zmserver.net, and I’ll keep the zacksmohawk.com site and email server etc. but I’m now looking at hosting my critical services on a VPS that won’t just get blatted by the hosting company.
Got an email yesterday (I think, I was out on my stag do) from those nice folks at 123-reg.com, informing me that “we ran a dodgy script, it deleted a shitload of your websites and data, really sorry to all those people who didn’t maintain their own backups, etc.”. Brilliant.
Luckily, I don’t really have a vast customer database, so it’s not catastrophic to my still-dormant company, but it’s certainly given me pause for thought.
I will need to (probably) manually re-upload and reconfigure the entire site/DB. Fairly annoying, to say the least.
Been looking at 1&1 or GoDaddy as potential service replacements. Will update as required.
This is the first fully auto-processed GridSquare. Not bad. That’s 785 islands processed in a couple of minutes, with zero failures, and when I attempt to re-process the whole area it produces zero new areas – proving that the process is not “harmful” to already processed coasts.
I’ve still got a little bit to do before I’m 100% happy with the process, including:
- Auto-remove all non-matched base land areas, as long as they are fully within the GridSquare
- Auto-select entire square upon successful processing
- Improve progress readout (remove System.outs)
- Improve end of process message dialog
- Include the total time taken by the process
- Tidy up and commit
That’s about it, really. Looking ahead, I might start to plan some sort of higher level of automation where we can set it off selecting and auto-matching GridSquares on its own, but for now I’m (nearly) happy with all this.
I know I said I was technically “code complete” with regards to the tasks remaining for base land map production, but I’ve taken a tiny detour (with a view to speeding up this remaining task): More automation.
Rather than individually pick every single minor island in each GridSquare, some of which I might actually miss, I’ve been working towards selecting the GridSquare itself and attempting to “auto-island” every coastal way, or set of ways, that form a closed loop on screen.
I’m having a few minor problems, but no big deal. Will have it fixed soon. After which, I will not be constrained by the time it takes to manually “auto-island” every tiny piece of rock that pops up above sea level.
The eventual aim is full automation, where the application can be set off auto-picking and processing GridSquares as required. Definitely a long way off this, though.
Will keep you informed.