Mirrors/coredns
1
0
Fork 0
mirror of https://github.com/coredns/coredns.git synced 2026-08-20 23:08:28 -04:00
Code Issues Packages Projects Releases Wiki Activity

plugin/cache: configure stale TTL and failure recheck (#8411)

Browse Source 
* plugin/cache: configure stale response TTL

Signed-off-by: houyuwushang <liuluoqianqiu@outlook.com>

* plugin/cache: delay stale refresh retries

Signed-off-by: houyuwushang <liuluoqianqiu@outlook.com>

* plugin/cache: validate stale refresh responses

Signed-off-by: houyuwushang <liuluoqianqiu@outlook.com>

---------

Signed-off-by: houyuwushang <liuluoqianqiu@outlook.com>
This commit is contained in:
houyuwushang
2026-08-14 14:27:42 +08:00
committed by GitHub
parent 6199dc50a5
commit 9520d725d6
8 changed files with 757 additions and 72 deletions
Download Patch File Download Diff File Expand all files Collapse all files

15
plugin/cache/README.md vendored
View File

@@ -40,7 +40,7 @@ cache [TTL] [ZONES...] {
success CAPACITY [TTL] [MINTTL]
denial CAPACITY [TTL] [MINTTL]
prefetch AMOUNT [[DURATION] [PERCENTAGE%]]
serve_stale [DURATION] [REFRESH_MODE [VERIFY_TIMEOUT]]
serve_stale [DURATION] [immediate [RESPONSE_TTL [FAILURE_RECHECK]] | verify [VERIFY_TIMEOUT [RESPONSE_TTL [FAILURE_RECHECK]]]]
servfail DURATION
disable success|denial [ZONES...]
keepttl
@@ -66,7 +66,7 @@ cache [TTL] [ZONES...] {
* `serve_stale`, when serve\_stale is set, cache will always serve an expired entry to a client if there is one
available as long as it has not been expired for longer than **DURATION** (default 1 hour). By default, the _cache_ plugin will
attempt to refresh the cache entry after sending the expired cache entry to the client. The
responses have a TTL of 0. **REFRESH_MODE** controls the timing of the expired cache entry refresh.
responses have a TTL of 0 by default for backward compatibility. **REFRESH_MODE** controls the timing of the expired cache entry refresh.
`verify` will first verify that an entry is still unavailable from the source before sending the expired entry to the client.
`immediate` will immediately send the expired entry to the client before
checking to see if the entry is available from the source. **REFRESH_MODE** defaults to `immediate`. Setting this
@@ -78,6 +78,17 @@ cache [TTL] [ZONES...] {
verify before falling back to the stale entry. The verify continues in the background and refreshes the
cache when it eventually succeeds, so subsequent queries see the fresh entry. The default of `0` means
wait until the upstream's own timeout (the original `verify` behavior). Example: `serve_stale 1h verify 100ms`.
**RESPONSE_TTL** sets the TTL returned with expired entries and defaults to `0`. RFC 8767 requires stale
responses to use a TTL greater than zero and recommends `30s`. In `immediate` mode it follows the mode,
for example `serve_stale 1h immediate 30s`. In `verify` mode it follows **VERIFY_TIMEOUT**, for example
`serve_stale 1h verify 100ms 30s`; use `0` as the timeout to wait for the upstream while setting a response
TTL, as in `serve_stale 1h verify 0 30s`. The response TTL must be a whole number of seconds.
**FAILURE_RECHECK** follows **RESPONSE_TTL** and limits how frequently a failed refresh is attempted again
for the same cache entry. While a refresh is in flight or its failure recheck period is active, the stale
entry is served immediately without another upstream request. A failed refresh leaves the stale cache entry
intact. The default of `0` preserves the existing retry behavior. RFC 8767 recommends `30s` and says this
value should not exceed 5 minutes. Examples: `serve_stale 1h immediate 30s 30s` and
`serve_stale 1h verify 100ms 30s 30s`.
* `servfail` cache SERVFAIL responses for **DURATION**. Setting **DURATION** to 0 will disable caching of SERVFAIL
responses. If this option is not set, SERVFAIL responses will be cached for 5 seconds. **DURATION** may not be
greater than 5 minutes.