Bug 678548

This is a followup to bug 620363

This code is somewhat intertwined with the fixes for bug 672725.  Additionally, since bug 678524 proves that qemu has a race condition with exec: processing, fixing libvirt will be inevitible if qemu rips exec: out, and good practice even if qemu leaves exec: support in.

I have removed the FutureFeature keyword and RFE from the BZ summary as this work is needed to address races in the migration code.

Requesting blocker for 6.1 if qemu bug 678524 is not fixed; if the qemu exec: bug is fixed then the libvirt changes of this bug can be deferred to 6.2.

Looks like bug 678524 will be fixed, and it's much less invasive.  Recommend deferring the libvirt changes to 6.2.

Upstream now has fd: support in 0.9.0 (modulo any fallout regression fixes that might still be necessary), as of the series ending in commit:

commit 15d757ac4e5d4be6972f05f33aa3fa85fbedd0bd
Author: Eric Blake <eblake>
Date:   Fri Mar 25 11:02:27 2011 -0600

    qemu: support fd: migration with compression
    
    Spawn the compressor ourselves, instead of requiring the shell.
    
    * src/qemu/qemu_migration.c (qemuMigrationToFile): Spawn
    compression helper process when needed.

This should automatically be picked up by rebasing in RHEL 6.2.

This should be fixed by the libvirt-0.9.2-1.el6 rebase

Verified this bug as PASS. Details please refer to the following:

Environment:
# uname -r
2.6.32-156.el6.x86_64
qemu-kvm-0.12.1.2-2.165.el6.x86_64
libvirt-0.9.2-1.el6.x86_64

Steps:
1. # virsh list --all
 Id Name                 State
----------------------------------
 15 rhel56               running

# virsh save rhel56 rhel56.save
Domain rhel56 saved to rhel56.save

2. From the log file 'getfd' and 'fd:migrate' can be found for domain save

# cat /etc/libvirt/libvirtd.conf
log_level = 1
log_outputs="1:file:/tmp/new"

# cat /tmp/new
...
19:49:13.337: 17266: debug : qemuMonitorMigrateToFd:1523 : mon=0xecf270 fd=22 flags=1
19:49:13.337: 17266: debug : qemuMonitorSendFileHandle:1920 : mon=0xecf270, fdname=migrate fd=22
...
19:49:13.337: 17266: debug : virJSONValueToString:1101 : result={"execute":"getfd","arguments":{"fdname":"migrate"},"id":"libvirt-5"}
19:49:13.337: 17266: debug : qemuMonitorJSONCommandWithFd:226 : Send command '{"execute":"getfd","arguments":{"fdname":"migrate"},"id":"libvirt-5"}' for write with FD 22
...
19:49:13.338: 17266: debug : virJSONValueToString:1101 : result={"execute":"migrate","arguments":{"detach":true,"blk":false,"inc":false,"uri":"fd:migrate"},"id":"libvirt-6"}
19:49:13.338: 17266: debug : qemuMonitorJSONCommandWithFd:226 : Send command '{"execute":"migrate","arguments":{"detach":true,"blk":false,"inc":false,"uri":"fd:migrate"},"id":"libvirt-6"}' for write with FD -1


-----
I can reproduce this bug with libvirt-0.8.7-18.el6.x86_64. It use exec: for domain save.

# virsh save rhel56 rhel56.save
error: Failed to save domain rhel56 to rhel56.save
error: operation failed: domain save job: unexpectedly failed

# cat /etc/libvirt/libvirtd.conf
log_level = 1
log_outputs="1:file:/tmp/old"

# cat /tmp/old
...
19:36:14.966: 14428: debug : qemuMonitorMigrateToFile:1412 : mon=0x7f79a0011af0 argv=0x7f79b75fc910 target=/root/rhel56.save offset=4096 flags=1
...
19:36:14.966: 14428: debug : virJSONValueToString:1062 : result={"execute":"migrate","arguments":{"detach":true,"blk":false,"inc":false,"uri":"exec:cat | { dd bs=4096 seek=1 if=/dev/null && dd bs=1048576; } 1<>'/root/rhel56.save'"}}
19:36:14.966: 14428: debug : qemuMonitorJSONCommandWithFd:218 : Send command '{"execute":"migrate","arguments":{"detach":true,"blk":false,"inc":false,"uri":"exec:cat | { dd bs=4096 seek=1 if=/dev/null && dd bs=1048576; } 1<>'/root/rhel56.save'"}}' for write with FD -1

Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory, and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

http://rhn.redhat.com/errata/RHBA-2011-1513.html