Hacker News Reader

Comment by NoZebra120vClip | Hacker News Reader

You know about Takeout, right?

If it is just email, it is more ergonomic to setup a local POP client like Thunderbird to download the emails. You don't have to use it to send email, just configure once and bring it up regularly to fetch the emails. I have a full copy of my gmail mailbox this way (already moved banking/govt services to fastmail anyway).

Yeah, this is what I do for email as well. Lets you seamlessly transition your backup to another service if/when you stop using gmail too. Though I use imap rather than pop, so that I don't end up with duplicates or other weirdness. (I keep incremental backups of the drive where the email is stored, so I'm not worried about a situation where both the cloud and backup copies get wiped.)

Here's the difference between IMAP and POP3. IMAP is a cloud service client, and POP3 is a store-and-forward, more or less. When you configure IMAP, you're operating on files on the server, and some or all of them may be cached, but the client will want to synchronize state with the server at all times. With POP3 you can specify "Leave on server" or "Delete from server". The POP client simply connects once and downloads everything (or everything that's new) and then you definitely have local copies of all those messages.

POP3 is good for use cases where you're rapidly migrating out and you want to wipe the messages clean, server-side. Or, if you believe the server will play hijinks with your messages or folders in some way, POP3 will give you a snapshot from time of download.

Have you ever tried to takeout a large amount of data across many services? Describing it as a process you do slowly is entirely fair.

For me, I just used google takeout to generate a giant 18gig zip and then I downloaded it. I do that once a year. It takes a long time to download, but it’s actually fairly quick and I wouldn’t describe it as a slow process.

For me there are always a bunch of services that fail or are missing some data. So you need to keep retrying.

With google takeout? I’ve never experienced that and subsequent downloads always yield the same results.

It's been about a year since my last round (so it is probably time to do another backup) but there were a few services that were flaky. It wasn't even me auditing for missing data, the takeout page would notify you that some data failed to be packaged up.

As a counter-example, I haven't had an entirely successful Google Takeout export in at least a couple of years, using their service to schedule the exports automatically every two months.

I always have a failure with Google Fit data, which reports 'Service failed to retrieve this item' on the same JSON file every time. I assume this is something corrupted at their end.

It's not that uncommon for my exports to intermittently show failures with other services -- for example, my latest export, taken on April 20th, also failed to include one of my YouTube videos, with the same 'Service failed to retrieve this item' error. That video is usually included successfully, so I'm guessing this was a glitch.

Nothing major, but I can well believe that others also experience regular errors, although I'm sure we're in the minority.

Combined or not, google keep notes without titles always get failed or errors in takeout html.

Where/how do you backup your takeout data?

A synology NAS and aws glacier.

Why does it matter if it's slow or fast? Because it's easy, regardless.

You just click the boxes for the services you want, it e-mails you when it's ready minutes/hours later, and you download the file(s) at the speed of your internet connection.

Considering it has to zip up many gigabytes of data from various sources, it works at an entirely reasonable speed.

The idea that it's too slow to be of value is a reason not to use it makes no sense.

Do you just not make backups of your data at all, because backups take hours?

You can auth it to your Dropbox account if you have one and they will export to it on a cadence into the /Apps folder.

Reply on news.ycombinator.com