Docker-selenium: Firefox 53 crashes immediately after initialization with 3.4.0-bismuth

[testphreak]

Meta -

Image(s):
All firefox images.

Image Version:
3.4.0-bismuth

Docker version:
17.04.0-ce, build 4845c56

OS:
Ubuntu 16.04.2 LTS

Expected Behavior -

Firefox 53 gets initialized and runs test

Actual Behavior -

Firefox 53 crashes after launch.

Info -

Relevant Hub log -

17:46:53.053 INFO - Got a request to create a new session: Capabilities [{marionette=true, acceptInsecureCerts=true, browserName=firefox, version=, platform=ANY}]
17:46:53.054 INFO - Trying to create a new session on test slot {seleniumProtocol=WebDriver, browserName=firefox, maxInstances=5, version=53.0, applicationName=, platform=LINUX}

Relevant node log -

23/04/2017 17:46:28 Got connection from client 10.1.214.174
23/04/2017 17:46:28   other clients:
23/04/2017 17:46:28 Normal socket connection
23/04/2017 17:46:28 Disabled X server key autorepeat.
23/04/2017 17:46:28   to force back on run: 'xset r on' (3 times)
23/04/2017 17:46:28 incr accepted_client=1 for 10.1.214.174:60266  sock=12
23/04/2017 17:46:28 Client Protocol Version 3.8
23/04/2017 17:46:28 Protocol version sent 3.8, using 3.8
23/04/2017 17:46:28 rfbProcessClientSecurityType: executing handler for type 2
23/04/2017 17:46:29 created   xdamage object: 0x20002f
23/04/2017 17:46:29 copy_tiles: allocating first_line at size 44
23/04/2017 17:46:37 rfbProcessClientNormalMessage: ignoring unsupported encoding type Enc(0x00000016)
23/04/2017 17:46:37 rfbProcessClientNormalMessage: ignoring unsupported encoding type Enc(0x00000015)
23/04/2017 17:46:37 rfbProcessClientNormalMessage: ignoring unsupported encoding type Enc(0x0000000F)
23/04/2017 17:46:37 rfbProcessClientNormalMessage: ignoring unsupported encoding type Enc(0xFFFFFEC6)
23/04/2017 17:46:37 Enabling full-color cursor updates for client 10.1.214.174
23/04/2017 17:46:37 Enabling NewFBSize protocol extension for client 10.1.214.174
23/04/2017 17:46:37 Using ZRLE encoding for client 10.1.214.174
23/04/2017 17:46:37 Pixel format for client 10.1.214.174:
23/04/2017 17:46:37   8 bpp, depth 8
23/04/2017 17:46:37   uses a colour map (not true colour).
23/04/2017 17:46:37 client useCopyRect: 10.1.214.174 -1
23/04/2017 17:46:37 client_set_net: 10.1.214.174  0.0013
23/04/2017 17:46:37 created selwin: 0x200030
23/04/2017 17:46:37 called initialize_xfixes()
23/04/2017 17:46:37 rfbProcessClientNormalMessage: ignoring unsupported encoding type Enc(0x00000016)
23/04/2017 17:46:37 rfbProcessClientNormalMessage: ignoring unsupported encoding type Enc(0x00000015)
23/04/2017 17:46:37 rfbProcessClientNormalMessage: ignoring unsupported encoding type Enc(0x0000000F)
23/04/2017 17:46:37 rfbProcessClientNormalMessage: ignoring unsupported encoding type Enc(0xFFFFFEC6)
23/04/2017 17:46:37 Enabling full-color cursor updates for client 10.1.214.174
23/04/2017 17:46:37 Enabling NewFBSize protocol extension for client 10.1.214.174
23/04/2017 17:46:37 Switching from ZRLE to raw Encoding for client 10.1.214.174
23/04/2017 17:46:38 rfbProcessClientNormalMessage: ignoring unsupported encoding type Enc(0x00000016)
23/04/2017 17:46:38 rfbProcessClientNormalMessage: ignoring unsupported encoding type Enc(0x00000015)
23/04/2017 17:46:38 rfbProcessClientNormalMessage: ignoring unsupported encoding type Enc(0x0000000F)
23/04/2017 17:46:38 rfbProcessClientNormalMessage: ignoring unsupported encoding type Enc(0xFFFFFEC6)
23/04/2017 17:46:38 Enabling full-color cursor updates for client 10.1.214.174
23/04/2017 17:46:38 Enabling NewFBSize protocol extension for client 10.1.214.174
23/04/2017 17:46:38 Switching from raw to ZRLE Encoding for client 10.1.214.174
17:46:53.065 INFO - Executing: [new session: Capabilities [{marionette=true, acceptInsecureCerts=true, browserName=firefox, version=, platform=ANY}]])
17:46:53.077 INFO - Creating a new session for Capabilities [{marionette=true, acceptInsecureCerts=true, browserName=firefox, version=, platform=ANY}]
1492969613132   geckodriver INFO    Listening on 127.0.0.1:3902
1492969613226   geckodriver::marionette INFO    Starting browser /opt/firefox-53.0/firefox with args ["-marionette"]
Failed to read: session.screen0.titlebar.left
Setting default value
Failed to read: session.screen0.titlebar.right
Setting default value
1492969614883   Marionette  INFO    Listening on port 44210
1492969614982   Marionette  WARN    TLS certificate errors will be ignored for this session
17:46:55.192 INFO - Detected dialect: W3C
17:46:55.207 INFO - Done: [new session: Capabilities [{marionette=true, acceptInsecureCerts=true, browserName=firefox, version=, platform=ANY}]]
17:46:55.367 INFO - Executing: [maximise window])
17:46:55.374 INFO - Done: [maximise window]
17:46:55.458 INFO - Executing: [implicit wait: 5000])
17:46:55.464 INFO - Done: [implicit wait: 5000]
17:46:55.544 INFO - Executing: [page load wait: 60000])
17:46:55.549 INFO - Done: [page load wait: 60000]
17:46:55.626 INFO - Executing: [script wait: 30000])
17:46:55.631 INFO - Done: [script wait: 30000]
17:46:55.708 INFO - Executing: [get: https://www.google.com])
17:46:56.781 INFO - Done: [get: https://www.google.com]
17:46:56.862 INFO - Executing: [get current url])
17:46:56.869 INFO - Done: [get current url]
17:46:57.042 INFO - Executing: [take screenshot])
17:46:57.085 INFO - Done: [take screenshot]
17:46:57.176 INFO - Executing: [delete session: e7da9232-7118-4809-aaab-c591ef27932e])
1492969617179   Marionette  INFO    New connections will no longer be accepted
[Child 383] WARNING: pipe error (3): Connection reset by peer: file /home/worker/workspace/build/src/ipc/chromium/src/chrome/common/ipc_channel_posix.cc, line 346
17:46:57.366 INFO - Done: [delete session: e7da9232-7118-4809-aaab-c591ef27932e]

My docker-compose.yml -

selhub:
  image: selenium/hub:3.4.0-bismuth
  environment:
    - GRID_MAX_SESSION=3
  ports:
    - 4444:4444

nodefirefoxdebug:
  image: selenium/node-firefox-debug:3.4.0-bismuth
  privileged: true
  environment:
    - NODE_MAX_INSTANCES=3
    - NODE_MAX_SESSION=3
    - DBUS_SESSION_BUS_ADDRESS=/dev/null
  volumes:
    - /dev/shm:/dev/shm
  ports:
    - 5900
  links:
    - selhub:hub

[manoj9788]

I just used the docker-compose.yml above and tried a sample test and VNC'd into it - it works. Maybe let's wait for someone else to reproduce this error.

[diemol]

It also works for me, but I can imagine that we are running different types of tests to check this. @testphreak it would be cool if you can provide some code snippet to reproduce it.

[timonag]

Going to this google page almost always kills FF

Docker: version 17.03.1-ce, build c6d412e
Host: Win10

node_1  | 15:46:13.101 INFO - Executing: [get: https://trends.google.com/trends/hottrends])
node_1  | 15:46:13.101 DEBUG - Executing: get [7dd6006c-a1ac-4a1f-a979-de93d2755db8, get {url=https://trends.google.com/trends/hottrends}]
node_1  | 15:46:13.103 DEBUG - CookieSpec selected: default
node_1  | 15:46:13.103 DEBUG - Auth cache not set in the context
node_1  | 15:46:13.103 DEBUG - Connection request: [route: {}->http://localhost:11211][total kept alive: 0; route allocated: 0 of 2000; total allocated: 0 of 2000]
node_1  | 15:46:13.104 DEBUG - Connection leased: [id: 2][route: {}->http://localhost:11211][total kept alive: 0; route allocated: 1 of 2000; total allocated: 1 of 2000]
node_1  | 15:46:13.104 DEBUG - Opening connection {}->http://localhost:11211
node_1  | 15:46:13.104 DEBUG - Connecting to localhost/127.0.0.1:11211
node_1  | 15:46:13.106 DEBUG - Connection established 127.0.0.1:35994<->127.0.0.1:11211
node_1  | 15:46:13.106 DEBUG - http-outgoing-2: set socket timeout to 10800000
node_1  | 15:46:13.107 DEBUG - Executing request POST /session/7dd6006c-a1ac-4a1f-a979-de93d2755db8/url HTTP/1.1
node_1  | 15:46:13.107 DEBUG - Target auth state: UNCHALLENGED
node_1  | 15:46:13.108 DEBUG - Proxy auth state: UNCHALLENGED
node_1  | 15:46:13.109 DEBUG - http-outgoing-2 >> POST /session/7dd6006c-a1ac-4a1f-a979-de93d2755db8/url HTTP/1.1
node_1  | 15:46:13.109 DEBUG - http-outgoing-2 >> Content-Type: application/json; charset=utf-8
node_1  | 15:46:13.109 DEBUG - http-outgoing-2 >> Content-Length: 52
node_1  | 15:46:13.110 DEBUG - http-outgoing-2 >> Host: localhost:11211
node_1  | 15:46:13.111 DEBUG - http-outgoing-2 >> Connection: Keep-Alive
node_1  | 15:46:13.111 DEBUG - http-outgoing-2 >> User-Agent: Apache-HttpClient/4.5.3 (Java/1.8.0_121)
node_1  | 15:46:13.112 DEBUG - http-outgoing-2 >> Accept-Encoding: gzip,deflate
node_1  | 15:46:13.113 DEBUG - http-outgoing-2 >> "POST /session/7dd6006c-a1ac-4a1f-a979-de93d2755db8/url HTTP/1.1[\r][\n]"
node_1  | 15:46:13.114 DEBUG - http-outgoing-2 >> "Content-Type: application/json; charset=utf-8[\r][\n]"
node_1  | 15:46:13.114 DEBUG - http-outgoing-2 >> "Content-Length: 52[\r][\n]"
node_1  | 15:46:13.115 DEBUG - http-outgoing-2 >> "Host: localhost:11211[\r][\n]"
node_1  | 15:46:13.115 DEBUG - http-outgoing-2 >> "Connection: Keep-Alive[\r][\n]"
node_1  | 15:46:13.117 DEBUG - http-outgoing-2 >> "User-Agent: Apache-HttpClient/4.5.3 (Java/1.8.0_121)[\r][\n]"
node_1  | 15:46:13.117 DEBUG - http-outgoing-2 >> "Accept-Encoding: gzip,deflate[\r][\n]"
node_1  | 15:46:13.118 DEBUG - http-outgoing-2 >> "[\r][\n]"
node_1  | 15:46:13.118 DEBUG - http-outgoing-2 >> "{"url":"https://trends.google.com/trends/hottrends"}"
node_1  | [Parent 90] WARNING: pipe error (52): Connection reset by peer: file /home/worker/workspace/build/src/ipc/chromium/src/chrome/common/ipc_channel_posix.cc, line 346
node_1  | [Parent 90] WARNING: pipe error (43): Connection reset by peer: file /home/worker/workspace/build/src/ipc/chromium/src/chrome/common/ipc_channel_posix.cc, line 346
node_1  |
node_1  | ###!!! [Parent][MessageChannel] Error: (msgtype=0x2C0085,name=PBrowser::Msg_Destroy) Channel error: cannot send/recv

[manoj9788]

Can you please try the same test in FF with the conventional setup of Grid with the jar file and put your findings here?

[testphreak]

@manoj9788, @diemol figured out the problem. By default GRID_BROWSER_TIMEOUT on the hub is set to 0 seconds. Increased the browser and grid timeouts value and all my tests ran successfully.

[timonag]

Well, conventional grid on Win10 (and, as reported, on Mac) did not reproduce the error. Grid on docker with this composition

version: "2"
services:
 hub:
  image: selenium/hub:3.4.0-bismuth
  ports:
   - 4444:4444
  environment:
   - no_proxy=localhost
   - GRID_TIMEOUT=60
   - GRID_BROWSER_TIMEOUT=60
   - GRID_DEBUG=true
 node1:
  image: selenium/node-firefox-debug:3.4.0-bismuth
  ports:
    - 5900:5900
  links:
   - hub
  environment:
   - HUB_PORT_4444_TCP_ADDR=hub
   - HUB_PORT_4444_TCP_PORT=4444

still fails at random point of the test.

[testphreak]

@timonag what does the no_proxy=localhost do? Otherwise looks fine to me. Let me know if you'd like me to keep issue open or you can create a new one.

[timonag]

Nothing helps, even a roll back to older versions of node. Will wait for FF54.