Dan Fandrich:

> > Seems like some kind of race condition?
> 
> Could be. Try running just that test but with the --repeat option to run
> it multiple times in a row, or with that and -j to run multiple copies in
> parallel, and I'm guessing it will fail at some point even on its own.

Indeed, when I run it with --repeat, the first run succeeds and the
second one fails.  Every time.  I assume this is a problem in the
test itself rather than in curl?

$ make test TFLAGS="--repeat=10 1542"
[...]
test 1542...[connection reuse with CURLOPT_MAXLIFETIME_CONN]
--p---oe--- OK (1   out of 10 , remaining: 00:27, took 3.061s, duration: 00:03)
test 1542...[connection reuse with CURLOPT_MAXLIFETIME_CONN]

 1542: output (log/stderr1542) FAILED:
--- log/check-expected  Sun Jun  8 21:09:11 2025
+++ log/check-generated Sun Jun  8 21:09:11 2025
@@ -1,5 +1,4 @@
 == Info: Connection #0 to host 127.0.0.1 left intact[CR][LF]
 == Info: Connection #0 to host 127.0.0.1 left intact[CR][LF]
 == Info: Connection #0 to host 127.0.0.1 left intact[CR][LF]
-== Info: shutting down connection #0[CR][LF]
-== Info: Connection #1 to host 127.0.0.1 left intact[CR][LF]
+== Info: Connection #0 to host 127.0.0.1 left intact[CR][LF]

 - abort tests
[...]

-- 
Christian "naddy" Weisgerber                          [email protected]
-- 
Unsubscribe: https://lists.haxx.se/mailman/listinfo/curl-library
Etiquette:   https://curl.se/mail/etiquette.html

Reply via email to