OpenStreetMap.org currently offline. Operations Team are working to restore - 15 December 2024 (Updated)

[Peter_Elderson]

@Peter_Elderson

I have a considerable ‘forward log’, I hope to get it uploaded soon!

Uploaded a ton of work I had compiled in JOSM; had two minor conflicts which were easily resolved.

[Firefishy]

We now have the new ISP also enabled and configured in Dublin. Old ISP still has higher priority for outbound traffic, but inbound works via either simultaneously (we use simple policy routing on host and “virtual router” on router to achieve this).

We will switch DNS when comfortable.

[chris66]

@Ferrocarriles_de_México

What font face is this?

I think this is the Default Win11 console font: Cascadia Mono.

[SekeRob]

Exactly: Cascadia Code Monospace.

GitHub - microsoft/cascadia-code: This is a fun, new monospaced font that includes...

[Makeena]

@Minh_Nguyen

iD does save your progress in the browser’s local storage. However, this storage has a finite size. (A red banner appears when you exceed the quota.) The mechanism to retrieve these stored changes can also be finicky. In some browsers, closing the tab like normal would actually cause iD to think you’ve intentionally discarded your changes, so you won’t get the option to restore them. Instead, you’d have to force-quit your browser or tab, or cause the browser or tab to crash. There are also some known issues where certain kinds of edit conflicts can cause iD to spin forever.

While it’s a good feature in general, if you can also save out an osmChange file as a backup, that would give you some peace of mind. Unfortunately, it is true that the option to review and back up your changes is gated behind a successful login session. So hopefully now that the site is back up, you can export those changes before rolling the dice on an upload.

I think I haven’t encountered this feature but the issues you describe are a reason why I would still vote for manual control over local cache saving instead of good luck and dirty tricks. Save the current state when I want it and don’t give the editor or browser the right to decide when to delete it, that’s a right reserved for the user.

I personally haven’t lost an edit that way but a comment. The browser reloaded the tab because it was running out of RAM and the comment text wasn’t restored. This wasn’t very bad in my case but it just shows an issue. The same might happen when switching tabs or apps/programs before finishing the comment, not just during a locked state.

[Minh_Nguyen (Minh Nguyễn)]

@Makeena

The browser reloaded the tab because it was running out of RAM and the comment text wasn’t restored.

Changeset tags probably aren’t stored locally like other state. Feel free to open an issue about that. Unfortunately, an osmChange file won’t persist this information either, but there’s a proposal to change that:

Proposal to include changeset tags in the .osc file format

I have created a proposal that affects the osmChange file format. If you are a software developer, or if you use this file format, then you may be interested in this technical proposal. Everyone is invited to comment on the wiki page. Please, cross post this announcement on the tagging mailing list on my behalf by sending an email to: tagging@openstreetmap.org

[Mammi71 (One feature, Six mappers and still More ways to map it)]

Danke an das Team!

[muralito]

@admin

Alternatively we could manually sync over a 4G + VPN link, but we have deemed this too high risk for the moment.

Always keep in mind what Andy said long long time ago.

Never underestimate the bandwidth of a station wagon full of tapes hurtling down the highway.

[Firefishy]

It was considered. 5 hours to Amsterdam by train for me and then more hours to get to Dublin, but I get the point.

lockheed sitting near the database servers has 6x 20TB disks

[Firefishy]

I have started producing the post-mortem: Produce post-mortem for HE.net outage in Amsterdam - 15 December 2024 · Issue #1196 · openstreetmap/operations · GitHub

Once completed it will be added to the https://operations.osmfoundation.org/ site.

[ZeLonewolf (Brian M Sperlongano)]

Does our new ISP have safeguards in place that would prevent the outage that we saw at HE.net?

[Firefishy]

Yes. We have 2 devices which each have their own 10G fibre uplink to confirmed separate remote side equipment, unlike we discovered with HE.net where both links went to the same remote side equipment. The links are “diverse” aka they follow different physical paths. The Equinix Internet routers run VRRP and we connect to the virtual IP addresses offered by these devices.

Equinix Internet also has multiple upstream peers and transits. Recently with HE.net we had issues where connectivity with HE.net and an AWS region were unreliable because of an issue with HE.net & AWS peering. HE.net were surprisingly not able to route the traffic via other transit routes to the affected AWS region.

Next time I am physically on-site we will test the equipment to ensure a disconnected link triggers the link failover as expected with alarms/alerts being triggered as expected.

[9u2]

On what do you base your expectation of uptime and resilience? Does your OSM subscription contain any SLA? Guarantees or warranties?

Last I checked OSM was entirely a volunteer project. For something this big to be online, at all, for any duration, is impressive, and something I’m personally thankful for. Merry Christmas!