Use async name resolving to improve net thread responsiveness #4421
Conversation
In the LookupIntern(), things changed are: 1. Call getaddrinfo_a() instead of getaddrinfo() if available, the former is a sync version of the latter; 2. Try using inet_pton()/inet_addr() to convert the input text to a network addr structure at first, if success the extra name resolving thread inside getaddrinfo_a() could be avoided; 3. An interruption point added in the waiting loop for return from getaddrinfo_a(), which completes the improve for thread responsiveness. A easy way to see the effect is to kick off a 'bitcoind stop' immediately after 'bitcoind -daemon', before the change it would take several, or even tens of, minutes on a bad network situation to wait for the running bitcoind to exit, now it costs only seconds. Signed-off-by: Huang Le <4tarhl@gmail.com>
|
Automatic sanity-testing: PASSED, see http://jenkins.bluematt.me/pull-tester/p4421_caf6150e9785da408f1e603ae70eae25b5202d98/ for binaries and test log. |
|
Untested ACK. The current behavior is very annoying, and this looks like a pretty clear improvement with very little added complexity. Nit: really the headers should be checked as well in configure, but realistically the lib checks should work on all reasonable platforms. |
|
Agreed @theuni BTW to rebase a pull you do not need to resubmit it. You can re-push to the same branch on github to update the current pull. |
|
@laanwj Yes I know that, I re-submit since I feel it is "cleaner" when just one commit included in one PR:) |
|
You still don't need to resubmit for that. If you rebase and force-push (push -f) you'll only have the commits that exist in your remote repository. |
|
Anyhow the code changes look good to me, going to test it. |
|
@laanwj Got it. Will try that next time... |
|
I've tested this for a week. No apparent problems. ACK. |
caf6150 Use async name resolving to improve net thread responsiveness (Huang Le)
| #endif | ||
|
|
||
| #ifdef HAVE_INET_PTON | ||
| #include <arpa/inet.h> |
There was a problem hiding this comment.
gitian-win fails to build on this see #4473
|
Reverting in #9229 due to glibc bugs in getaddrinfo_a. |
NOTE: This is a re-submit of PR 4259, just re-base to latest master for avoiding merge conflct, no real change was made. The previous discussion could be checked here: #4259.
In the LookupIntern(), things changed are:
A easy way to see the effect is to kick off a 'bitcoind stop' immediately after 'bitcoind -daemon', before the change it would take several, or even tens of, minutes on a bad network situation to wait for the running bitcoind to exit, now it costs only seconds.
Signed-off-by: Huang Le 4tarhl@gmail.com