rpm: produce sane amount of logging when creating a rootfs
PATCH 1/1 rpm: produce sane amount of logging when creating a rootfs
Previously there was lots of irrelevant noise in the logs, because we also wanted to be able to debug postinst issues easily. I have adjusted the logging levels so that postinst info is still written to the logs, but other things are not.
[YOCTO #13119]
--- meta/lib/oe/package_manager.py | 2 +- ...ge-logging-level-around-scriptlets-t.patch | 55 +++++++++++++++++++ meta/recipes-devtools/rpm/rpm_4.14.2.1.bb | 1 + 3 files changed, 57 insertions(+), 1 deletion(-) create mode 100644 meta/recipes-devtools/rpm/files/0016-rpmscript.c-change-logging-level-around-scriptlets-t.patch
meta/lib/oe/package_manager.py +1 -1
@@ -966,7 +966,7 @@ class RpmPM(PackageManager): os.environ['RPM_ETCCONFIGDIR'] = self.target_rootfs dnf_cmd = bb.utils.which(os.getenv('PATH'), "dnf") - standard_dnf_args = ["-v", "--rpmverbosity=debug", "-y", + standard_dnf_args = ["-v", "--rpmverbosity=info", "-y", "-c", oe.path.join(self.target_rootfs, "etc/dnf/dnf.conf"), "--setopt=reposdir=%s" %(oe.path.join(self.target_rootfs, "etc/yum.repos.d")), "--installroot=%s" % (self.target_rootfs),
meta/recipes-devtools/rpm/files/0016-rpmscript.c-change-logging-level-around-scriptlets-t.patch new file +55 -0
55 lines not on this page · raw message
meta/recipes-devtools/rpm/rpm_4.14.2.1.bb +1 -0
@@ -40,6 +40,7 @@ SRC_URI = "git://github.com/rpm-software-management/rpm;branch=rpm-4.14.x \ file://0004-build-pack.c-remove-static-local-variables-from-buil.patch \ file://0001-perl-disable-auto-reqs.patch \ file://0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch \ + file://0016-rpmscript.c-change-logging-level-around-scriptlets-t.patch \ " PE = "1"
Signature · 1 line
2.17.1
I'm slightly worried that this will break on-target RPM users. Since info is commonly enabled during on-target installations to track which packages have been installed and what their progress is.
Has it been test that this does not break dnf on the target?
--Mark
Quoted 103 lines · Alexander Kanavin
Previously there was lots of irrelevant noise in the logs, because we also wanted to be able to debug postinst issues easily. I have adjusted the logging levels so that postinst info is still written to the logs, but other things are not.
[YOCTO #13119]
Signed-off-by Alexander Kanavin <[email protected]>--- meta/lib/oe/package_manager.py | 2 +- ...ge-logging-level-around-scriptlets-t.patch | 55 +++++++++++++++++++ meta/recipes-devtools/rpm/rpm_4.14.2.1.bb | 1 + 3 files changed, 57 insertions(+), 1 deletion(-) create mode 100644 meta/recipes-devtools/rpm/files/0016-rpmscript.c-change-logging-level-around-scriptlets-t.patchmeta/lib/oe/package_manager.py +1
-1@@ -966,7 +966,7 @@ class RpmPM(PackageManager): os.environ['RPM_ETCCONFIGDIR'] = self.target_rootfs dnf_cmd = bb.utils.which(os.getenv('PATH'), "dnf") - standard_dnf_args = ["-v", "--rpmverbosity=debug", "-y", + standard_dnf_args = ["-v", "--rpmverbosity=info", "-y", "-c", oe.path.join(self.target_rootfs, "etc/dnf/dnf.conf"), "--setopt=reposdir=%s" %(oe.path.join(self.target_rootfs, "etc/yum.repos.d")), "--installroot=%s" % (self.target_rootfs),meta/recipes-devtools/rpm/files/0016-rpmscript.c-change-logging-level-around-scriptlets-t.patch new file +55
-0@@ -0,0 +1,55 @@ +From 989e425d416474c191b020d0825895e3df4bd033 Mon Sep 17 00:00:00 2001 +From: Alexander Kanavin <[email protected]> +Date: Thu, 10 Jan 2019 18:14:18 +0100 +Subject: [PATCH] rpmscript.c: change logging level around scriptlets to INFO + from DEBUG + +That way we can debug scriptlet failures without writing lots of +irrelevant noise to rootfs logs. + +Upstream-Status: Inappropriate [oe-core specific] +Signed-off-by: Alexander Kanavin <[email protected]> +--- + lib/rpmscript.c | 8 ++++---- + 1 file changed, 4 insertions(+), 4 deletions(-) + +diff --git a/lib/rpmscript.c b/lib/rpmscript.c +index 2b0e43862..e319673f1 100644 +--- a/lib/rpmscript.c ++++ b/lib/rpmscript.c +@@ -226,7 +226,7 @@ static char * writeScript(const char *cmd, const char *script) + if (Ferror(fd)) + goto exit; + +- if (rpmIsDebug() && (rstreq(cmd, "/bin/sh") || rstreq(cmd, "/bin/bash"))) { ++ if (rpmIsVerbose() && (rstreq(cmd, "/bin/sh") || rstreq(cmd, "/bin/bash"))) { + static const char set_x[] = "set -x\n"; + /* Assume failures will be caught by the write below */ + Fwrite(set_x, sizeof(set_x[0]), sizeof(set_x)-1, fd); +@@ -258,7 +258,7 @@ static rpmRC runExtScript(rpmPlugins plugins, ARGV_const_t prefixes, + char *mline = NULL; + rpmRC rc = RPMRC_FAIL; + +- rpmlog(RPMLOG_DEBUG, "%s: scriptlet start\n", sname); ++ rpmlog(RPMLOG_INFO, "%s: scriptlet start\n", sname); + + if (script) { + fn = writeScript(*argvp[0], script); +@@ -310,7 +310,7 @@ static rpmRC runExtScript(rpmPlugins plugins, ARGV_const_t prefixes, + sname, strerror(errno)); + goto exit; + } else if (pid == 0) {/* Child */ +- rpmlog(RPMLOG_DEBUG, "%s: execv(%s) pid %d\n", ++ rpmlog(RPMLOG_INFO, "%s: execv(%s) pid %d\n", + sname, *argvp[0], (unsigned)getpid()); + + fclose(in); +@@ -353,7 +353,7 @@ static rpmRC runExtScript(rpmPlugins plugins, ARGV_const_t prefixes, + reaped = waitpid(pid, &status, 0); + } while (reaped == -1 && errno == EINTR); + +- rpmlog(RPMLOG_DEBUG, "%s: waitpid(%d) rc %d status %x\n", ++ rpmlog(RPMLOG_INFO, "%s: waitpid(%d) rc %d status %x\n", + sname, (unsigned)pid, (unsigned)reaped, status); + + if (reaped < 0) {meta/recipes-devtools/rpm/rpm_4.14.2.1.bb +1
-0@@ -40,6 +40,7 @@ SRC_URI = "git://github.com/rpm-software-management/rpm;branch=rpm-4.14.x \ file://0004-build-pack.c-remove-static-local-variables-from-buil.patch \ file://0001-perl-disable-auto-reqs.patch \ file://0001-rpm-rpmio.c-restrict-virtual-memory-usage-if-limit-s.patch \ + file://0016-rpmscript.c-change-logging-level-around-scriptlets-t.patch \ " PE = "1"
Quoted 7 lines · Mark Hatle
I'm slightly worried that this will break on-target RPM users. Since info is
commonly enabled during on-target installations to track which packages have
been installed and what their progress is.Has it been test that this does not break dnf on the target?
Easy fix: make this specific to class-native.
That's probably sensible even if it doesn't break dnf, because then target RPM behaves as usual.
Ross
Quoted 10 lines · Ross
Quoted 5 lines
I'm slightly worried that this will break on-target RPM users. Since info is commonly enabled during on-target installations to track which packages have been installed and what their progress is.
Has it been test that this does not break dnf on the target?
Easy fix: make this specific to class-native.
That's probably sensible even if it doesn't break dnf, because then target RPM behaves as usual.
I'm not too worried as dnf interacts with rpm via rpm's python API; what rpm prints out should be of no consequence.
The extra lines are probably useful for the target as well, since postinsts can fail in various ways there too.
If this passes the dnf/rpm tests on the AB, I'd leave it like it is.
Alex