Polarian's Website

Running an XMPP Sprint, a review of London XMPP Sprint 2026

From Friday 6th November to Sunday 8th November, 8 XMPP community members met for an XMPP Sprint in Greenwich, London.

I was the co-organiser for the event, helping the organiser, Matthew Fennell, with the outreach and formalities, such as the wiki page and blogposts.

For more information about the sprint, please read the wiki page. This blogpost will mainly cover what it is like organising a sprint, where it went well, and where it fell apart.

Organisation

Picking a location for a sprint

This is probably one of the most fundamental choices which need to be made when organising a sprint. Major cities have the best (and usually cheapest) links to international airports, along with comprehensive public transport networks making it easier for people to attend, however all of this comes with a much higher cost.

I originally tried to find a University to sponsor the sprint, the space is the most expensive part of running a sprint, and without a sponsor it was bound to be expensive. I contacted my former University asking if they would provide the space, however I never recieved a response and I didn't want to push it that much. A friend of mine found a lecturer at Kings College London (KCL) which had written many papers on XMPP, so I reached out to him, however he had no interest in helping with a sprint.

This left us with no other choice but try to find an affordable venue, something which was next to impossible in London. Many of the places we found wanted £400-500/day, and renting meeting rooms in the library for a long period of time would add up to a similar.

Matthew then found a venue in Greenwich which fit the budget, and was a central location which had good connections to most airports. This then became the venue for the sprint.

Dates

Planning started at the beginning of 2026, Matthew knew after FOSDEM he wanted to run an XMPP Sprint and I was interested in helping out as I was also in London.

We decided on running it in November close to the end of the year as it gave plenty of time to plan and organise the sprint.

Most sprints seem to do Friday, Saturday and Sunday to maximise the amount of time while also keeping the amount of holiday time required to attend to a couple of days for the attendees.

Wiki

Now the venue and dates were confirmed, we had to ensure it was all documented. This is where the mediawiki instance hosted by the XSF infrastructure team comes in.

I mainly wrote the wiki page as I had more time than Matthew to handle it. There is a standard format for the wiki page along with information on organising a sprint on the Sprints wiki page.

Then you make a subpage to the wiki page, copy the format and fill in the details for your sprint. Should be simple enough even if the organiser hasn't used mediawiki before.

Outreach

Now the important part, outreach. Typically posting it on XMPP to the XSF discussions channel, and also to the sprints MUC are good ways to get it out there.

Posting it to the members mailing list, or getting a member to post it there is another good place to pick up attendees, as many of major developers are XSF members.

Blogpost

Next, it is time to write a blogpost. Firstly fork the XMPP website repository on Github.

Then clone your fork and add a new entry to content/blog. You can then compose the blogpost entry in markdown, for an example you can check the PR I submitted for London XMPP Sprint.

Adding it to the events section of the XMPP website

After you finished composing your blogpost, before you commit, you should also add an entry to the data/events.json file to add the information about your sprint. An example entry can be seen yet again on the PR I submitted for London XMPP Sprint.

Finally you can commit, and push the changes to your fork, then you can make a PR to the main repository.

Shortly after your PR is merged you should see your blog post on the XMPP website blog and your event entry on the XMPP website's events page.

Visiting the venue

If you haven't visited the venue yet, it is a good time to do so. What is most important is two fundamental requirements for hacking, decent internet connection and a lot of power sockets.

Power

Look around for how many power sockets at the venue, if there is under 2 sockets per person, or these sockets are too far away from the table to work at, then extension leads/power strips will be needed to connect more devices. General rule of thumb is to have 2 sockets per person (as people typically like to charge their phone and their laptop at the same time), some people might even have two laptops (Matthew Fennell for example :p).

Internet

In order for developers to write code, they usually need internet. This should be dual stack if possible (IPv4 and IPv6) and have no egress filtering (outbound ports which are blocked, or sites which are blocked).

Nice to haves:

London XMPP Sprint Review

The good

Space

The venue had plenty of space, and Matthew (Fennell) had plenty of power strips to connect up all the attendee devices. Power was not an issue.

Catering

The venue served a variety of drinks, both alcoholic and non-alcoholic, and food, which from the general vibe given off from the attendees, were both great.

Connectivity/Internet

Venue had decent wifi, however it was locked behind a captive portal with some egress filtering which was especially annoying for me because I was working on servers with arbitrary ports. The wifi was also single stack, and FreeBSD (which I daily drive on my laptop) has pretty bad support for wifi making ethernet preferable.

I had worked hard on getting a network stack which would relay the wifi at the venue and then tunnel it to a trusted network, which would be able to provide dual stack networking. However stuff happened, more on that in the unlucky section.

Attendees

In total 8 people attended the sprint, out of a capacity of 15, that was over half the venue filled.

The bad

Miscommunication

There was miscommunication between the venue and Matthew (Fennell) which led to the venue costing more than expected. It thankfully was still within budget, but it did mean the great deal we thought we had originally ended up being about double what was initially anticipated.

Thank you to Matthew Fennell which not only paid for the expected amount, but also covered the unexpected costs of the venue, which ended up being covered by free Breakfast and Lunch along with drinks all weekend. People still paid for their dinner however.

Infrastructure work the night before

I had to work on my infrastructure the night before the sprint, which meant I was up almost the entire night swapping out a router and a network switch within my infrastructure, which had to be done late at night due to the downtime it would cause.

This meant on day 1 (Friday) I was exhausted, and barely able to function. I did not get much work done, and by the afternoon I passed out.

I feel horrible for this because not only is it rude, and sets a bad example, but it wasted time I could have been spending productively. Unfortunately this was not the only mess up of the sprint.

Missing day 2 (Saturday)

After being knackered the day before, and getting back rather late because I had a 1.5 hr commute to and from the venue in Greenwich (longer than any other attendee), I passed out and woke up the afternoon on Saturday, missing half the sprint by the time I even woke up, and by the time I managed to get down to the venue with delays and the long travel distance, it was already almost 6pm and there was little time to work on anything.

Not only is this yet again rude, considering I was one of the co-organisers, but I thought I had messaged Matthew Fennell letting him know that I was going to be late, but instead I was so wiped out from sleep depravation that it was all imagined, and it was only until I got to the venue and checked my messages I realised I never sent the messages I thought I did, I hadn't even composed them!

I felt awful for leaving Matthew wondering where I could have been, and for just being absent for the whole day.

I guess there is two morals of this story, the first one is no matter what is wrong, duct tape a solution together and roll with it, do not do infrastructure maintenance the night before an event, especially a 3 day sprint.

The other moral of the story is, there is a reason people use cloud hosting, or co-locate their hardware. On-premise comes with commitments which can lead to unfavourable work hours, and outright bad luck for when you need to do maintenance. The benefits of on-premise to me still outweigh the negatives, however for anyone who is into infrastructure, this is a major consideration and compromise you will have to make when you decide where you plan to have your physical hardware.

The unlucky

Networking gear failing on day 1

After a lot of work to get the networking stack working, it all failed on the morning of day 1, with no energy or time to work on it after a night of working on my infrastructure, I left it at home and relied on the captive portal wifi at the venue.

The issue was the device I was using to relay the wifi failed to bridge to my OpenBSD router, and this hardware failure couldn't be remedied quickly as I didn't have another device already setup which was capable of relaying wifi, or providing an LTE connection to the router.

Dual stack failure on day 3

On day 3, OpenBSD refused to stay connected to the wireguard endpoint, it would connect and work then disconnect 2 mins later, with limited time mid-sprint I didn't have much time to diagnose the issue, so abandoned dual stack connectivity.

I bridged an LTE modem to the OpenBSD router, plugged the OpenBSD router into a network switch, which then had my laptop and wifi AP broadcasting the "XMPPSprint" SSID for attendees. It worked, no egress filtering either due to it being on LTE, unfortunately still IPv4 only.

No worries, I tunnelled my devices over wireguard for IPv6 connectivity which I needed to connect to certain non-public regions of my infrastructure.

LTE modem failure

In the afternoon, after the setup working all morning flawlessly, there was a slight annoyance with the speed. When I had viewed the venue I found the LTE speed was good and it should be much faster than it was. So I decided to move the LTE modem over to where I did the original test a few months before, however upon unplugging the ethernet from the LTE modem, it died, and I lost £70 worth of gear randomly.

This was the second dose of pure bad luck, that a embedded device which usually lasts a decade, just randomly died despite being relatively new.

After the sprint I realised the reason the speeds were not fast, despite me testing them before, was because I picked up the wrong sim in a rush on the Sunday morning, I was meant to be using a different one which would have been far faster.

Nevertheless, slightly slow network without filtering would have still been better than egress filtering, and I disappointed the hobby project I put so much effort into ended up failing, twice. But this was just pure bad luck!

Conclusion

The Sprint from the sounds of things was enjoyed by all, and a massive success.

To me there is mixed feelings, I enjoyed discussing XMPP with people much smarter than myself, I enjoyed hanging out after the sprint for dinner, I learned some new things about the XMPP internals, and I met some new people, or at least, got to know people I hadn't really spoken to too much before.

I was disappointed about the failure of my gear, it was completely random and there was no way I would have known or been able to anticipate it.

I am angry at myself for the failure on my part. I pass out on day 1, I miss day 2 and only actually ended up enjoying the sprint like I should have on day 3, all because of bad decision making I made the night before the first day. The sprint occurred just a week or two after me fighting off serious sickness, where I had to cancel XMPP OpMeet, an online conference I had moved to the end of October and then ultimately bailed on being too unwell. Already having disappointed many from this, and having Matthew Fennell needing to step in my place for a FOSS meetup I organised while I was in hospital during my sickness left a bad feeling. Then coming back and failing this badly was a poor way to get back into the swing of things.

I apologise to the attendees for my poor decisions leading up to the sprint, and for the pure bad luck.

The most important thing I, and anyone who is reading this should take from this experience, is nothing is more important than you getting a good nights sleep before an event, everything else can wait. I thought I could power through it, but underestimate exhaustion or attempt to fight it too long, and your body simply passes out on its own accord.

Thank you for reading!