You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
This is what the response latency looks like when hitting infra.cid.contact:
And this is the same CID lookup, except pointed at cid.contact:
Investigate why this is happening.
cid.contact landing serves a static web UI that is hosted on GitHub Pages, with a specially crafted iframe handling at go ingress. As a result, the find response latency from cid.contact can vary widely depending on how long DNS resolution.To fix this we should use built-in s3 support in cloudfront and point the landing to the web UI generated and uploaded to s3 somewhere with cloudfront routing config to go with it. Use this opportunity to investigate removing the built-in iframe from storetheindex binary such that changing any aspect of Web UI it does not require rebuilding storetheindex.
The text was updated successfully, but these errors were encountered:
As a result, the find response latency from cid.contact can vary widely depending on DNS resolution
how is the find request/response impacted by the web ux? that's a direct query to the cid.contact server and should be orthogonal to the latencies/dns lookups around loading the initial web page, right?
You are right; apologies for the late night brain fart.
cid.contact DNS record rightly points to cloudfront. Right now I don't know why the response time varies when hitting cid.contect directly.
digsometimes returns different IPs for cid.contact vs. infra.cid.contact for example which i am not sure why:
$ dig infra.cid.contact
; <<>> DiG 9.10.6 <<>> infra.cid.contact
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 11920
;; flags: qr rd ra; QUERY: 1, ANSWER: 5, AUTHORITY: 0, ADDITIONAL: 1
;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 1232
;; QUESTION SECTION:
;infra.cid.contact. IN A
;; ANSWER SECTION:
infra.cid.contact. 257 IN CNAME d3gfvti28lf5wu.cloudfront.net.
d3gfvti28lf5wu.cloudfront.net. 17 IN A 143.204.176.13
d3gfvti28lf5wu.cloudfront.net. 17 IN A 143.204.176.66
d3gfvti28lf5wu.cloudfront.net. 17 IN A 143.204.176.74
d3gfvti28lf5wu.cloudfront.net. 17 IN A 143.204.176.95
;; Query time: 139 msec
;; SERVER: 10.1.0.1#53(10.1.0.1)
;; WHEN: Wed Dec 21 09:15:28 GMT 2022
;; MSG SIZE rcvd: 153
$ dig cid.contact
; <<>> DiG 9.10.6 <<>> cid.contact
;; global options: +cmd
;; Got answer:
;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 10724
;; flags: qr rd ra; QUERY: 1, ANSWER: 4, AUTHORITY: 0, ADDITIONAL: 1
;; OPT PSEUDOSECTION:
; EDNS: version: 0, flags:; udp: 1232
;; QUESTION SECTION:
;cid.contact. IN A
;; ANSWER SECTION:
cid.contact. 53 IN A 13.224.132.85
cid.contact. 53 IN A 13.224.132.100
cid.contact. 53 IN A 13.224.132.12
cid.contact. 53 IN A 13.224.132.55
;; Query time: 145 msec
;; SERVER: 10.1.0.1#53(10.1.0.1)
;; WHEN: Wed Dec 21 09:15:17 GMT 2022
;; MSG SIZE rcvd: 104
masih
changed the title
Point cid.contact CNAME directly to CloudFront for stable response latency
Response latency from cid.contact varies widely compared to other CNAMEs
Dec 21, 2022
This is what the response latency looks like when hitting
infra.cid.contact
:And this is the same CID lookup, except pointed at
cid.contact
:Investigate why this is happening.
cid.contact
landing serves a static web UI that is hosted on GitHub Pages, with a specially crafted iframe handling at go ingress. As a result, the find response latency from cid.contact can vary widely depending on how long DNS resolution.To fix this we should use built-in s3 support in cloudfront and point the landing to the web UI generated and uploaded to s3 somewhere with cloudfront routing config to go with it. Use this opportunity to investigate removing the built-in iframe fromstoretheindex
binary such that changing any aspect of Web UI it does not require rebuildingstoretheindex
.The text was updated successfully, but these errors were encountered: