Open
Description
What happened?
Suddenly when we run some automation jobs the grid hub queue size is growing and not allocating sessions though we have enough sessions. We had to restart the grid to overcome this issue. This happens very frequently. We are using podman for configuring our grid. This seems relevant to this issue #1865. Tests will run for some time, and suddenly this issue is popping up.
Capabilities:
browserName: 'chrome',
acceptInsecureCerts: true,
'goog:chromeOptions': {
prefs: {
intl: {accept_languages: "en-US"},
},
w3c: true,
args: [
'incognito',
'--disable-browser-side-navigation',
'--ignore-certificate-errors',
'--no-sandbox', '--start-fullscreen',
'--disable-infobars',
'--disable-dev-shm-usage'
]
}
Command used to start Selenium Grid with Docker (or Kubernetes)
podman run -d --name selenium-hub \
--network myprivate \
--cpus 2 \
-p 4442:4442 \
-p 4443:4443 \
-p 4444:4444 \
--restart always \
--memory 20000M \
--memory-reservation 19000M \
-e GRID_BROWSER_TIMEOUT=300 \
-e GRID_TIMEOUT=300 \
-e GRID_MAX_SESSION=60 \
-e GRID_MAX_INSTANCES=3 \
-e GRID_CLEAN_UP_CYCLE=5000 \
-e GRID_UNREGISTER_IF_STILL_DOWN_AFTER=180000 \
-e GRID_NEW_SESSION_WAIT_TIMEOUT=60 \
docker.io/selenium/hub:4.14.1-20231025
# Set the number of nodes
NUM_NODES=150
NETWORK_NAME=private
MEMORY_LIMIT=4000M
MEMORY_RESERVATION=3900M
for ((i=1; i<=$NUM_NODES; i++)); do
NODE_NAME="selenium-chrome-node-$i"
podman run -d --name $NODE_NAME \
--network $NETWORK_NAME \
--cpus 1.5 \Relevant log output
NAOperating System
Red Hat Enterprise Linux release 8.9 (Ootpa)
Docker Selenium version (image tag)
4.14.1-20231025
Selenium Grid chart version (chart version)
NA