git/list[1] front-page[2] threads[3] people[4] search[5] about
 

[PATCHv3 01/10] git p4 test: wait longer for p4d to start and test its pid

From
PWPete Wyckoff <pw@padd.com>
Date
Jun 28, 2012, 02:48 UTC
Message-ID
<20120628024807.GA16349@padd.com>
In-Reply-To
<7vfw9g78w5.fsf@alter.siamese.dyndns.org>

Running tests at high parallelism on a slow machine, 5 sec is not enough to wait for p4d to start. Change it to 5 minutes, adding an environment variable P4D_START_PATIENCE to shrink that if needed in automated test environments.

Also check if the pid of the p4d that we started is still around. If not, quit waiting for it immediately.

Remove all the confusing && chaining and simplify the code.
Thanks-to: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Pete Wyckoff <pw@padd.com>
---
 t/lib-git-p4.sh | 38 ++++++++++++++++++++++++++++++++------
 1 file changed, 32 insertions(+), 6 deletions(-)
gitster@pobox.com wrote on Wed, 27 Jun 2012 14:16 -0700:
> In other words, I think it would make it much more clear to write
> this part like this:
[..]

Much nicer. I think I got hung up on always &&-chaining instead of writing the code so it works. This version steals your suggestions almost verbatim. Tested under scenarios where p4d was slow, hung or died, as well as when it works as usual.

Just this patch changed; if it's easy to replace one in the series? Else I could resend all 10.

		-- Pete
diff --git a/t/lib-git-p4.sh b/t/lib-git-p4.sh
index 121e380..eea46a4 100644
--- a/t/lib-git-p4.sh
+++ b/t/lib-git-p4.sh
@@ -33,17 +33,42 @@ pidfile="$TRASH_DIRECTORY/p4d.pid"
 
 start_p4d() {
 	mkdir -p "$db" "$cli" "$git" &&
+	rm -f "$pidfile" &&
 	(
 		p4d -q -r "$db" -p $P4DPORT &
 		echo $! >"$pidfile"
 	) &&
-	for i in 1 2 3 4 5 ; do
-		p4 info >/dev/null 2>&1 && break || true &&
-		echo waiting for p4d to start &&
+
+	# This gives p4d a long time to start up, as it can be
+	# quite slow depending on the machine.  Set this environment
+	# variable to something smaller to fail faster in, say,
+	# an automated test setup.  If the p4d process dies, that
+	# will be caught with the "kill -0" check below.
+	i=${P4D_START_PATIENCE:-300}
+	pid=$(cat "$pidfile")
+	ready=
+	while test $i -gt 0
+	do
+		# succeed when p4 client commands start to work
+		if p4 info >/dev/null 2>&1
+		then
+			ready=true
+			break
+		fi
+		# fail if p4d died
+		kill -0 $pid 2>/dev/null || break
+		echo waiting for p4d to start
 		sleep 1
-	done &&
-	# complain if it never started
-	p4 info >/dev/null &&
+		i=$(( $i - 1 ))
+	done
+
+	if test -z "$ready"
+	then
+		# p4d failed to start
+		return 1
+	fi
+
+	# build a client
 	(
 		cd "$cli" &&
 		p4 client -i <<-EOF
@@ -53,6 +78,7 @@ start_p4d() {
 		View: //depot/... //client/...
 		EOF
 	)
+	return 0
 }
 
 kill_p4d() {
-- 
1.7.11.1.69.g485c1e0
Previous: Junio C HamanoNext: Junio C Hamano
Message 4 of 14 in “[PATCHv2 00/10] git p4 test fixes”
  1. Pete WyckoffJun 27, 2012
  2. 01/10 git p4 test: wait longer for p4d to start and test its pidPete Wyckoff, Jun 27, 2012
  3. Junio C HamanoJun 27, 2012
  4. 01/10 git p4 test: wait longer for p4d to start and test its pidPete Wyckoff, Jun 28, 2012
  5. Junio C HamanoJun 28, 2012
  6. 02/10 git p4 test: use real_path to resolve p4 client symlinksPete Wyckoff, Jun 27, 2012
  7. 03/10 git p4 test: simplify quoting involving TRASH_DIRECTORYPete Wyckoff, Jun 27, 2012
  8. 04/10 git p4 test: never create default test repoPete Wyckoff, Jun 27, 2012
  9. 05/10 git p4 test: rename some "git-p4 command" stringsPete Wyckoff, Jun 27, 2012
  10. 06/10 git p4 test: check for error message in failed testPete Wyckoff, Jun 27, 2012
  11. 07/10 git p4 test: copy source indeterminatePete Wyckoff, Jun 27, 2012
  12. 08/10 git p4 test: cleanup_git should make a new $gitPete Wyckoff, Jun 27, 2012
  13. 09/10 git p4 test: split up big t9800 testPete Wyckoff, Jun 27, 2012
  14. 10/10 git p4 test: fix badp4dir testPete Wyckoff, Jun 27, 2012

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.