Log the real reason for why posix_fadvise failed

[TedLyngmo]

reclaimFilePageCache did not set errno but rdbSaveInternal which is logging the error assumed it did. This makes sure errno is set.

Fixes #429

[codecov]

Codecov Report

Attention: Patch coverage is 33.33333% with 2 lines in your changes are missing coverage. Please review.

Project coverage is 68.45%. Comparing base (5b1fd22) to head (babd006).

Additional details and impacted files

@@            Coverage Diff            @@
##           unstable     #430   +/-   ##
=========================================
  Coverage     68.44%   68.45%           
=========================================
  Files           109      109           
  Lines         61671    61673    +2     
=========================================
+ Hits          42209    42216    +7     
+ Misses        19462    19457    -5     

Files	Coverage Δ
src/util.c	69.49% <33.33%> (-0.22%)	⬇️

... and 11 files with indirect coverage changes

[madolson]

Makes sense to me, given https://man7.org/linux/man-pages/man2/posix_fadvise.2.html.

Did you encounter this issue while running in production or just reviewing code. This will influence whether or not we backport the fix.

[TedLyngmo]

Did you encounter this issue while running in production or just reviewing code. This will influence whether or not we backport the fix.

I noticed it in our test environment when I tried tracking down reasons for misc. "Resource temporarily unavailable" issues - so I don't really know the real reason for why posix_fadvise failed.

I also added a PR in Redis: redis/redis#13246

[madolson]

OK, sounds like enough reason to backport.

[TedLyngmo]

@madolson It looks like this didn't make it into 7.2.6, but I do see it in 8.0.0-rc1. Was that a mistake or how are backports released?

[madolson]

@TedLyngmo That was a mistake, we didn't have our backporting strategy sorted correctly. It will be in 7.2.7 which is getting released tomorrow.