Changes between Version 14 and Version 15 of TroubleShooting

Timestamp:

Aug 22, 2009, 1:56:03 AM (15 years ago)

Author:

Bruno Cornec

Comment:

finish the detailed example

TroubleShooting

@@ -147 +147 @@
 We suppose that you're at the prompt after the crash.
 Prepare on your NFS server the needed conten for debugging the case:
+
+First download both the normal and the debug mondo packages:
 {{{
 # cd /dir/exported/to/mondo
-# 
+# wget ftp://ftp.mondorescue.org/test/rhel/5/mondo-2.2.9-0.20090729004531.rhel5.x86_64.rpm
+# wget ftp://ftp.mondorescue.org/test/rhel/5/mondo-debuginfo-2.2.9-0.20090729004531.rhel5.x86_64.rpm
+}}}
+
+Then on the original platform make the backup the way you're used to using the downloaded mondo package (and mindi of course). On the same platform you also have to install the valgrind package.
+Then create a tar file containing the mondo debig info and valgrind content that you make available on your NFS server:
+{{{
+# rpm2cpio mondo-debuginfo-2.2.9-0.20090729004531.rhel5.x86_64.rpm | cpio -idum
+# tar cfz mondo.tgz ./usr
+# rm -rf usr
+# tar cfz valgrind.tgz /usr/bin/valgrind /usr/lib*/valgrind
+}}}
+
+Then on the restored client, at the prompt you can extract the content, and use it:
+{{{
+# cd /
+# tar xfz /tmp/isodir/valgrind.tgz
+# tar xfz /tmp/isodir/mondo.tgz
+# valgrind --log-file=/tmp/valg.log --show-reachable=yes --track-origins=yes --leak-check=full mondorestore -K 99 -Z interactive
+}}}
+
+Then send those files to the dev team with a picture of the crash:
+{{{
+/var/log/mondorestore.log
+/tmp/valg.log
+}}}
+
 
 === Attaching to Running Processes ===