New timeout nomenclature #3817
New timeout nomenclature #3817
Commits on Jun 27, 2022
-
Now that aliases show themselves as such to other tweaks, we'll need the other tweaks to know about this detail. Moving the tweak up will allow a new function visible to the others without a prior declaration. Refs f885637
-
tweak: Teach regular tweaks about aliases
The way we now pass aliases, we lose the original priv pointer. So we need to resolve the original parameter in turn. Fixes f885637
-
man: Extract varnish-params(7) from varnishd(1)
The latter is already long enough with command line options coverage alone, and with incoming additions to the parameters documentation now seems a good time for this extraction.
-
-
man: Document the new timeout nomenclature
Most timeouts don't comply, and will be migrated gradually with deprecated aliases to maintain compatibility. The resulting list of timeouts will also help us notice what we may be missing and offer a reliable naming convention for new timeouts.
-
vcc: Teach vcc_IdIs() to fall back to aliases
We could even change the return value to a string and return where p is when a match is found, or NULL. This way a call site could check whether the original identifier or an alias was found, and for example issue a warning.
-
param: Give aliases a proper default value
Otherwise it breaks the param.reset command.
-
param: Rename first_byte_timeout to beresp_start_timeout
Conversely, the VCL variable `bereq.first_byte_timeout` is replaced by `beresp.start_timeout`.
-
param: Rename between_bytes_timeout to beresp_idle_timeout
And the VCL variable `bereq.between_bytes_timeout` is in turn replaced by `beresp.idle_timeout`.
-
-
-
param: Rename connect_timeout to bereq_connect_timeout
And this time, no impact on VCL code, only VRT.
-
vtcp: New VTCP_set_send_timeout() function
Convenience for parity with VTCP_set_read_timeout().
-
param: Rename idle_send_timeout to resp_idle_interrupt
Exceptionally, one existing VTC was migrated to the new name immediately since it relied on debug logs containing 'idle_send_timeout'. Turning the timeout to an interrupt, I tried to update the parameter description to better convey what it does. This does not change the current behavior, but I tried to highlight what is wrong with this timeout and the mismatch between its purpose for HTTP/1 and h2. Depending on where we look at the documentation, between the parameter or VCL variables, we get a slightly different description. None of them are accurate, but the consensus seems to have always been the parameter description from before h2 or the 'sess.idle_send_timeout' variable. For session-level timeouts, we may consider an h2_send_timeout parameter in the future to distinguish between socket writes and underlying h2 streams timing out on a condvar. This h2_send_timeout parameter would map to SO_SNDTIMEO for h2 client connections. Refs varnishcache#2927
-
param: Rename pipe_timeout to pipe_idle_timeout
Note to self, absolutely no test coverage for this parameter.
-
param: Rename send_timeout to resp_send_timeout
Everything mentioned in the commit log where idle_send_timeout is renamed to resp_idle_timeout applies here as well.
-
param: Rename timeout_* parameters
They needed to be taken care of simultaneously to avoid hassles related to their current cohabitation with resp timeouts.
-
