Environment running Tomcat on SUSE12 OS is short on RAM

• Type: Bug
• Status: Open
• Resolution: Unresolved
• Affects Version/s: 7.1.X, 7.2.X
• Fix Version/s: None
• Component/s: Environments
• Labels:
  • Enviro-TF
  • QA-AUTOMATION
  • QA-FP
• QA Test Name:
  • LocalFile.CPBlogs#AddBlogsEntry
  • LocalFile.CPCustomfields#EditCustomFieldUserBoolean
  • LocalFile.CPDocumentsandmedia#EditDocument
  • LocalFile.CPMessageboards#AddMBThread
  • LocalFile.CPRoles#DeleteRegularRole
  • LocalFile.CPUsersandorganizations#EditUserSite
  • LocalFile.ContentPagesWithStaging#PublishEditedContentPage
  • LocalFile.Forms#PreviewAndPublishForm
  • LocalFile.FragmentsUI#AddFragmentToContentPage
  • LocalFile.OAuth2#addAndEditApplication
  • LocalFile.OAuth2#viewAndRevokeGrantedAuthorizations
  • LocalFile.PGStaging#ActivateDeactivateRemoteStaging
  • LocalFile.Permissions#ViewBlogsPortletAsGuestPG
  • LocalFile.PortalSmoke#Smoke
  • LocalFile.StagingUsecaseWithVersioning#StagingPublishLast
  • LocalFile.WebContentDisplay#AddWithStructure
  • LocalFile.WorkflowCalendar#ApproveEventWithInvitedUser
• QA Test Score: 83

Tests running on SUSE12

From the link above, we can see that sometimes the Error is reported on Testray as:

java.lang.Exception: Element is not present at "//span[contains(@class,'sign-in')]/a/span"

Other times it is more generic, as:

The build failed prior to running the test.

But the stated reason for shutdown for each is:

 [echo] /opt/dev/projects/github/liferay-portal-7.1.x-private/build-test.xml:4730: Unable to find startup message after 30 minutes.

So it looks like sometimes it tries to load localhost after the startup message, but the page can't load - other times the startup message isn't reached. In either case, it looks like this environment doesn't have enough memory to complete the build and startup within the allotted 30 minutes.

---

Here are the build params:

ANT_HOME:/opt/java/ant, ANT_OPTS:-Dhttp.nonProxyHosts="cloud*|files.liferay.org.es|it.liferay.com|localhost|mirrors-no-cache.lax.liferay.com|release*|test*" -Dhttp.proxyHost=squid.lax.liferay.com -Dhttp.proxyPort=3128 -server -verbose:gc -Xloggc:/tmp/ant-gc.log -Xms1024m -Xmx3072m -Xss2m -Xverify:none -XX:+DisableExplicitGC -XX:+HeapDumpOnOutOfMemoryError -XX:+PrintGCCause -XX:+PrintGCDetails -XX:+UseParallelOldGC -XX:HeapDumpPath=/tmp/ant-oom-heap-dump.bin -XX:MaxNewSize=512m -XX:MaxPermSize=512m -XX:MaxTenuringThreshold=0 -XX:NewSize=256m -XX:PermSize=100m -XX:SurvivorRatio=65536 -XX:TargetSurvivorRatio=0

Here are the current startup params (found in the console log):

-Xms2560m -Xmx2560m -XX:MaxNewSize=1536m -XX:MaxMetaspaceSize=512m -XX:MetaspaceSize=512m -XX:NewSize=1536m -XX:SurvivorRatio=7

And the SUSE 12 VMs are running on 8GB of RAM, when they should probably have 16:

[email protected]:~> free -m
             total       used       free     shared    buffers     cached
Mem:          8192       2642       5549          0          0       1437
-/+ buffers/cache:       1204       6987
Swap:         2048       1568        479