Taylor Marvin
@taylormar.vin
Software dev working in FinTech, primarily working with C# & PowerShell. Former systems & network admin.
I enjoy electronic music, computers, and time spent with my friends. Views are my own.
https://github.com/tsmarvin
I enjoy electronic music, computers, and time spent with my friends. Views are my own.
https://github.com/tsmarvin
oof, on top of having misconfigurations in my app - I was doubling down by having cloudflare send a solid number of those requests through the ai labrinth 😅
January 22, 2026 at 6:04 PM
oof, on top of having misconfigurations in my app - I was doubling down by having cloudflare send a solid number of those requests through the ai labrinth 😅
Looking at the numbers made me realize how bad it was, I hadn’t been concerned when I saw it initially because it’s all traffic on the same host, but woof... Definitely looks like I had a few things misconfigured. Should be a straightforward enough fix but that’s tomorrow’s problem.
January 22, 2026 at 7:39 AM
Looking at the numbers made me realize how bad it was, I hadn’t been concerned when I saw it initially because it’s all traffic on the same host, but woof... Definitely looks like I had a few things misconfigured. Should be a straightforward enough fix but that’s tomorrow’s problem.
Yeah. Between the refresh endpoint and the getsession endpoint I’m doing something wrong or something’s borked somewhere.
Its just a disproportion number of requests
Its just a disproportion number of requests
January 22, 2026 at 4:29 AM
Yeah. Between the refresh endpoint and the getsession endpoint I’m doing something wrong or something’s borked somewhere.
Its just a disproportion number of requests
Its just a disproportion number of requests
Nothing like trial by fire(hose) lol
On a tangentially related note - For some reason my music lookup/playlist sharing app has a ton of refresh events on a consistent cycle with the Idunno.atproto library that I need to debug at some point - but it certainly makes my requests graphs look neat
On a tangentially related note - For some reason my music lookup/playlist sharing app has a ton of refresh events on a consistent cycle with the Idunno.atproto library that I need to debug at some point - but it certainly makes my requests graphs look neat
January 22, 2026 at 4:23 AM
Nothing like trial by fire(hose) lol
On a tangentially related note - For some reason my music lookup/playlist sharing app has a ton of refresh events on a consistent cycle with the Idunno.atproto library that I need to debug at some point - but it certainly makes my requests graphs look neat
On a tangentially related note - For some reason my music lookup/playlist sharing app has a ton of refresh events on a consistent cycle with the Idunno.atproto library that I need to debug at some point - but it certainly makes my requests graphs look neat
Good luck!
Out of curiosity are you hosting it on personal machine?
If you are make sure to label the plug so nobody unplugs it to use the vacuum while you’re out [its happened to me multiple times for other services I host]
Out of curiosity are you hosting it on personal machine?
If you are make sure to label the plug so nobody unplugs it to use the vacuum while you’re out [its happened to me multiple times for other services I host]
January 22, 2026 at 4:09 AM
Good luck!
Out of curiosity are you hosting it on personal machine?
If you are make sure to label the plug so nobody unplugs it to use the vacuum while you’re out [its happened to me multiple times for other services I host]
Out of curiosity are you hosting it on personal machine?
If you are make sure to label the plug so nobody unplugs it to use the vacuum while you’re out [its happened to me multiple times for other services I host]
You’re gonna need to take read only fridays to a whole new level
January 22, 2026 at 4:02 AM
You’re gonna need to take read only fridays to a whole new level
I’m curious what the expensive part was? Compute or bandwidth?
January 20, 2026 at 5:17 PM
I’m curious what the expensive part was? Compute or bandwidth?
Unless you use k8s or ecs or the like, then you get Waay more granular back off and retry support
January 20, 2026 at 3:26 AM
Unless you use k8s or ecs or the like, then you get Waay more granular back off and retry support
Container all the things! Things like ECS and k8s make life a lot easier too, but really docker (or your preferred alternative) is pretty great out of the box. Just set your restart policy to automatic and it’ll essentially behave like a windows service, with slightly less crash off back out support
January 20, 2026 at 3:24 AM
Container all the things! Things like ECS and k8s make life a lot easier too, but really docker (or your preferred alternative) is pretty great out of the box. Just set your restart policy to automatic and it’ll essentially behave like a windows service, with slightly less crash off back out support
Seems cool, but like it’d be a reasonable amount of extra work for a not-super-common scenario. Could probably get most of the benefits by running things in k8s
January 20, 2026 at 2:19 AM
Seems cool, but like it’d be a reasonable amount of extra work for a not-super-common scenario. Could probably get most of the benefits by running things in k8s
Are you going to build your own solution for the container? Or just use something like pdsmoover?
January 20, 2026 at 1:30 AM
Are you going to build your own solution for the container? Or just use something like pdsmoover?
Clearly a valid expense, team morale is important
January 19, 2026 at 2:04 AM
Clearly a valid expense, team morale is important
In case it’s helpful in conjunction with the mouse tracking, I have a utility function for changing window focus on windows: gist.github.com/tsmarvin/3e9...
TM-WindowsUtility
TM-WindowsUtility. GitHub Gist: instantly share code, notes, and snippets.
gist.github.com
January 17, 2026 at 9:21 PM
In case it’s helpful in conjunction with the mouse tracking, I have a utility function for changing window focus on windows: gist.github.com/tsmarvin/3e9...
Thank you both for the explanation. That makes a lot more sense
January 17, 2026 at 8:21 PM
Thank you both for the explanation. That makes a lot more sense
Thank you, that helps a bit. So the concurrency error is really unrelated to the cid recalculation?
January 17, 2026 at 8:19 PM
Thank you, that helps a bit. So the concurrency error is really unrelated to the cid recalculation?
I’m definitely confusing myself this morning
January 17, 2026 at 8:05 PM
I’m definitely confusing myself this morning
Maybe I’m confusing things. What’s the advantage of keeping the cid the same?
I’m using PutRecord like an upsert, but I don’t mind people knowing the records are actually different if it’s edited
I’m using PutRecord like an upsert, but I don’t mind people knowing the records are actually different if it’s edited
January 17, 2026 at 8:01 PM
Maybe I’m confusing things. What’s the advantage of keeping the cid the same?
I’m using PutRecord like an upsert, but I don’t mind people knowing the records are actually different if it’s edited
I’m using PutRecord like an upsert, but I don’t mind people knowing the records are actually different if it’s edited
Is that effectively an edit? I’m making updates to existing rkeys to replace them but I think they end up with a new cid
January 17, 2026 at 7:39 PM
Is that effectively an edit? I’m making updates to existing rkeys to replace them but I think they end up with a new cid
Here’s my atproto implementation if you’d like to check it out at all:
github.com/tsmarvin/Bri...
github.com/tsmarvin/Bri...
BridgeBeats/src/Domain/Implementations/Services/ATProtoStorageService.cs at 4d90a70de6a15173d61d857420eeb0e24caf7cd4 · tsmarvin/BridgeBeats
Music is universal. Your links should be too. Contribute to tsmarvin/BridgeBeats development by creating an account on GitHub.
github.com
January 17, 2026 at 7:28 PM
Here’s my atproto implementation if you’d like to check it out at all:
github.com/tsmarvin/Bri...
github.com/tsmarvin/Bri...
Interesting - I’m not noticing a huge number of concurrency errors in any of my logging, but I implemented this before you added your recent lexicon updates and I haven’t made those updates yet. Would that being in the PDS logs or should I see it in my app?
January 17, 2026 at 7:28 PM
Interesting - I’m not noticing a huge number of concurrency errors in any of my logging, but I implemented this before you added your recent lexicon updates and I haven’t made those updates yet. Would that being in the PDS logs or should I see it in my app?
Then I logged in again to the web with the main password, and replaced it in the app which started working again. But I’ve now reverted back to the app password and it works fine now :/
January 17, 2026 at 7:23 PM
Then I logged in again to the web with the main password, and replaced it in the app which started working again. But I’ve now reverted back to the app password and it works fine now :/
I think it may have had something to do with age verification on the account actually. It seems the app password part might have been unrelated. I logged in with the app password to the web, that prompted me for age verification, which failed because apparently that can’t be set with an app password
January 17, 2026 at 7:23 PM
I think it may have had something to do with age verification on the account actually. It seems the app password part might have been unrelated. I logged in with the app password to the web, that prompted me for age verification, which failed because apparently that can’t be set with an app password
Though I did just start getting this weird error this morning:
January 17, 2026 at 7:07 PM
Though I did just start getting this weird error this morning: