Records.snap prevents proper queries after bad parent setup

[ezelkow1]

While looking in to another issue I hit this with master

I had a setup of one cache(8080), pointed at an nginx(80) origin on the same box which was setting various CC headers (public, no-cache, max-age some with multiple headers some on single lines). ATS would cache fine. Then I added the box as a parent of itself and tried various settings of go_direct, parent_is_proxy=false, and ignore_self_detect=true. At some point my queries would start returning a 503. (dest_domain=. parent="thisboxfqdn.com:80" various_params)

At that point even if I remove the line from the parent.config and restart ATS I would continue to get back 503's. This happens until I delete the records.snap file. So something in there is being cached/held incorrectly since if the parent line is removed to the original setup then you would expect it to work again as before especially after a restart

[github-actions[bot]]

This issue has been automatically marked as stale because it has not had recent activity. Marking it stale to flag it for further consideration by the community.

[bryancall]

@ezelkow1 Sound like it might be the information about the parent being down is recorded in the metrics that is in the records.snap file.