How to do distributed locking
Why I kept it
Ten years after this essay, I hit the same kind of problem building AI agents on tryalexandria.fr. On long-horizon tasks, with many tool calls and a lot of data access, each step has to happen once, even when part of the state sits on the client and I do not control delivery. I handle it with PGMQ on Supabase: a Postgres queue that is easy to run.
François Paupier
Martin Kleppmann's blog8 February 2016Public source
Read the article
In this reference
Martin Kleppmann's essay on distributed locking.
Cite the original work
Martin Kleppmann (8 February 2016). How to do distributed locking. Martin Kleppmann's blog. https://martin.kleppmann.com/2016/02/08/how-to-do-distributed-locking.html
View BibTeX
@misc{how_to_do_distributed_locking,
author = {Martin Kleppmann},
title = {{How to do distributed locking}},
howpublished = {Martin Kleppmann's blog},
url = {https://martin.kleppmann.com/2016/02/08/how-to-do-distributed-locking.html},
year = {2016}
}
