Skip to main content
11 events
when toggle format what by license comment
Oct 9, 2021 at 0:00 history tweeted twitter.com/StackSoftEng/status/1446626515513401346
Oct 7, 2021 at 19:01 history edited JamesBoyZ CC BY-SA 4.0
added 124 characters in body
Oct 7, 2021 at 18:26 history edited JamesBoyZ CC BY-SA 4.0
added 248 characters in body
Oct 7, 2021 at 18:09 history edited JamesBoyZ CC BY-SA 4.0
edited title
Oct 7, 2021 at 18:03 history edited JamesBoyZ CC BY-SA 4.0
added 879 characters in body
Oct 7, 2021 at 17:03 history edited JamesBoyZ CC BY-SA 4.0
added 879 characters in body
Oct 7, 2021 at 13:47 comment added Ewan I guess I worry when you start to try and fix something that your underlying system is supposed to do. If the write is always supposed to work but doesnt, will your remedy be any more reliable? Rather than write a fall back for "can't write to db for obscure reason" you might be better writing a "flip to Azure when AWS is down" for those times when something major goes down
Oct 7, 2021 at 13:37 comment added JamesBoyZ @Ewan that's a good question. Actually, I think I made the wrong statement. We do have a retry process in place to consume from the queue. However, how long some event remains stuck in the queue entirely depends on how long we have trouble with DynamoDB. If Amazon does a good job, it should be short. I took a look at 1 year statistics, we have less than 300 events ever arrived at the fallback queue. Chance of failure is low but we want to use the same system for more services internally going forward. So I'm looking for a way to improve our promise to the clients.
Oct 7, 2021 at 12:56 answer added Ewan timeline score: 1
Oct 7, 2021 at 12:38 comment added Ewan how short is very short? could you just delay the response a few ms?
Oct 7, 2021 at 11:40 history asked JamesBoyZ CC BY-SA 4.0