starred/openwrt-docker
1
0
Fork 0
mirror of https://github.com/AlbrechtL/openwrt-docker.git synced 2026-10-27 16:01:01 +00:00
Code Issues Packages Projects Releases Wiki Activity

Fix endless wait in qemu_qmp.sh when a qemu-ga status response is lost

Browse Source 
The qga socket is shared by several clients (healthcheck, web backend,
after-boot command script). qemu-ga reports "exited": true only once and
drops the PID afterwards, so if that response is lost every further
guest-exec-status fails and 'qemu_qmp.sh -c' loops forever. This made
run_command_after_openwrt_boot.sh never detect the OpenWrt boot and
test_openwrt_after_boot_cmd fail sporadically.

Abort on error responses and add a timeout to the boot detection poll.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
Albrecht Lohofener
2026-10-23 20:47:15 +02:00
co-authored by Claude Opus 5.5
parent 5af163b06f
commit 495c4ea70b
2 changed files with 8 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/qemu_qmp.sh
+6
View File
@@ -79,6 +79,12 @@ while getopts ":hvrqsSVRc" option; do
while true; do # Wait for command exit
RETURN_JSON=`echo '{"execute": "guest-exec-status", "arguments": { "pid": '${PID}'}}' | nc -w 1 -U /run/qga.sock`
#echo >&2 "RETURN_JSON: $RETURN_JSON"
# qemu-ga drops the PID after reporting "exited": true once. If that response got lost (e.g. received by
# another client of the shared socket), every further status request fails and we would wait forever.
if echo "$RETURN_JSON" | grep -q '"error"'; then
echo >&2 "Command error: guest-exec-status failed! RETURN_JSON: $RETURN_JSON"
exit 1
fi
export EXITCODE=`echo $RETURN_JSON | sed -n 's/.*"exitcode": \([0-9]*\).*/\1/p'`
export OUT_DATA=`echo $RETURN_JSON | sed -n 's/.*"out-data": "\([^"]*\).*/\1/p' | base64 -d -w 0`
export ERR_DATA=`echo $RETURN_JSON | sed -n 's/.*"err-data": "\([^"]*\).*/\1/p' | base64 -d -w 0`
src/run_command_after_openwrt_boot.sh
+2 -1
View File
@@ -36,7 +36,8 @@ fi
set +e # Enable that the next command can fail
while true; do
# Check if OpenWrt is booted
/run/qemu_qmp.sh -c 'logread | grep "procd: - init complete"' >/dev/null 2>&1
# Use a timeout to never get stuck in a single poll
timeout 30 /run/qemu_qmp.sh -c 'logread | grep "procd: - init complete"' >/dev/null 2>&1
if [ $? -eq 0 ]; then
echo "Detected OpenWrt is booted"
break