Weird bug with debmirror + libpam_ldap ?
Hi!
I'm hit by a strange bug, and I really don't know to wich package
report, nor how to further debug it.
Mi personal machine runs sid and I have an i386 Debian mirror,
updated dayly with debmirror in a cron job. At 2003/04/21 debmirror
stopped working, with the logs showing nothing. This machine has
it's accounts imported from an ldap server via libpam-ldap, and also
uses libnss-ldap to resolve nss services.
Using --debug option, I geti (non-us mirror):
----------------------------------------------------------------
/mirrors/scripts$ debmirror /mirrors/debian-non-us/ -h non-us.debian.org -r /debian-non-US --nosource --dist=woody,sarge,sid -s non-US/main,non-US/contrib,non-US/non-free -p --debug
/usr/bin/debmirror: Mirroring to /mirrors/debian-non-us/ from ftp://anonymous:non-us.debian.org//debian-non-US/
/usr/bin/debmirror: Arches: i386
/usr/bin/debmirror: Dists: woody,sarge,sid
/usr/bin/debmirror: Sections: non-US/main,non-US/contrib,non-US/non-free
Net::FTP: Net::FTP(2.65)
Net::FTP: Exporter(5.566)
Net::FTP: Net::Cmd(2.21)
Net::FTP: IO::Socket::INET(1.26)
Net::FTP: IO::Socket(1.27)
Net::FTP: IO::Handle(1.21)
Net::FTP=GLOB(0x8486618)<<< 220 klecker.debian.org FTP server (vsftpd)
Net::FTP=GLOB(0x8486618)>>> user anonymous
Net::FTP=GLOB(0x8486618)<<< 331 Please specify the password.
/mirrors/scripts$
----------------------------------------------------------------
At the same time, in /var/log/messages I get:
----------------------------------------------------------------
Apr 24 17:21:30 pcdaniel perl: nss_ldap: reconnecting to LDAP server...
Apr 24 17:21:30 pcdaniel perl: nss_ldap: reconnected to LDAP server after 1 attempt(s)
----------------------------------------------------------------
Doing a strace, the last lines are (editted out only the "relevant"):
----------------------------------------------------------------
socket(PF_UNIX, SOCK_DGRAM, 0) = 4
connect(4, {sa_family=AF_UNIX, path="/dev/log"}, 16) = 0
send(4, "<14>Apr 24 17:10:00 perl: nss_ld"..., 66, 0) = 66
socket(PF_INET, SOCK_STREAM, IPPROTO_IP) = 5
setsockopt(5, SOL_TCP, TCP_NODELAY, [1], 4) = 0
connect(5, {sa_family=AF_INET, sin_port=htons(389), sin_addr=inet_addr("192.168.0.2")}, 16) = -1 EINPROGRESS (Operation now in progress)
write(5, "0\f\2\1\1`\7\2\1\3\4\0\200\0", 14) = 14
read(5, "0\f\2\1\1a\7\n\1", 9) = 9
read(5, "\0\4\0\4\0", 5) = 5
write(5, "0\201\275\2\1\2c\201\267\4\ro=Aplik, c=cl\n\1\2\n\1\0\2"..., 192) = 192
read(5, "0\201\204\2\1\2d\177\4", 9) = 9
read(5, "!uid=daniel,ou=People,o=Aplik,c="..., 126) = 126
read(5, "0\f\2\1\2e\7\n\1", 9) = 9
read(5, "\0\4\0\4\0", 5) = 5
send(4, "<14>Apr 24 17:10:00 perl: nss_ld"..., 81, 0) = 81
rt_sigaction(SIGPIPE, {SIG_DFL}, NULL, 8) = 0
time([1051218600]) = 1051218600
rt_sigprocmask(SIG_SETMASK, [RTMIN], NULL, 8) = 0
--- SIGPIPE (Broken pipe) @ 0 (0) ---
+++ killed by SIGPIPE +++
----------------------------------------------------------------
So, it's like perl is trying to connect directly to the LDAP server
(other programs like "ls" use nscd) and in some way it chokes.
As this started happening at April 21, I revised the upgrade logs (I
upgrade all work days, and this was monday, so the opgrade list is large):
----------------------------------------------------------------
ldap-utils libldap2 libldap2-dev ntop
abiword abiword-common adduser alsa-base alsa-headers alsa-source alsaconf
base-config base-passwd blas blas-dev boa ca-certificates cpp-3.2 cupsys-bsd
cupsys-client cvs cvs-buildpackage ddd debconf debconf-doc debconf-utils
debootstrap developers-reference distcc e2fsck-static e2fsprogs fastjar
fontconfig g++-3.2 g77-3.2 galeon gcc-3.0-base gcc-3.2 gcc-3.2-base
gcc-3.2-doc gcc-3.3-base gcj-3.2 gdeskcal gij-3.2 glibc-doc gnome-gv grace
groff groff-base gtkglarea5 gtkglarea5-dev gtkterm gzip java-common kdevelop
kdevelop-data kdevelop-doc kernel-package libatk1.0-0 libatk1.0-data
libatk1.0-dev libatk1.0-doc libblkid1 libc6 libc6-dev libcupsys2
libcupsys2-dev libffi2 libfontconfig1 libfontconfig1-dev libfribidi0 libg2c0
libgail-common libgail-dev libgail17 libgcc1 libgcj-common libgcj3
libgcj3-dev libgtkmm2.0-1c102 libgtkmm2.0-dev libgtkmm2.0-doc liblua40
liblualib40 libmetacity0 libssl-dev libssl0.9.6 libssl0.9.7 libstdc++3
libstdc++5 libstdc++5-3.3-doc libstdc++5-dev libstdc++5-doc libstlport4.5c102
libwxgtk2.4 libxine-dev libxine1 libxml2 libxml2-dev lintian locales menu
metacity mime-support nano net-tools nscd openoffice.org openoffice.org-bin
openoffice.org-help-en openoffice.org-l10n-en openoffice.org-l10n-es openssl
pike7 pike7-crypto pike7-gdbm pike7-gl pike7-gtk pike7-gz pike7-image
pike7-mysql pike7-pg pike7.2 pike7.2-crypto pike7.2-gdbm pike7.2-gl
pike7.2-gtk pike7.2-gz pike7.2-image pike7.2-mysql pike7.2-pg pike7.4
pike7.4-core pike7.4-gdbm pike7.4-gtk pike7.4-image pike7.4-manual pike7.4-pg
pike7.4-svg powermgmt-base prc-tools-doc prc-tools-m68k prc-tools-utils
python-numeric python2.1-numeric python2.2-numeric quick-reference-en
setserial slang1 slang1-dev sudo tcc ttf-dustin tv-fonts v4l-conf xine-ui
----------------------------------------------------------------
I think that the culprits are libldap2 or nscd, but I don't realy
know perl nor libnet-perl, so I'm realy lost :-(
Thanks for your help!
Daniel.
Reply to: