QA/Browser Technologies/Services/Releases/SyncServer 08102011

Information

• Train: Sync Server 08-10-2011

Links

• Bugs/Issues: see Bugzilla
• Test Plans
  • https://wiki.mozilla.org/QA/Sync/Test_Plan#Servers
  • https://wiki.mozilla.org/QA/Sync/Test_Plan/Test_Level_Summaries#Smoketests_plus_2
• Stage Documentation: https://intranet.mozilla.org/QA/Server_Weekly_Trains_Staging
• Stage Environment
  • https://intranet.mozilla.org/Services/Ops/StageCluster
  • https://intranet.mozilla.org/QA/Mozilla_Services_Environment

Deployment Ticket/Maintenance Window Info

• bug 676109 - Deploy storage server 1.10-1 to staging
• bug 670946 - Release train bug for server-core 2.0
• bug 676351 - Deploy sreg server rpm-1.0-1 to production
• https://intranet.mozilla.org/Services/Ops/ChangeWindow_20110817
• https://intranet.mozilla.org/Services/Ops/ChangeWindow_20110818

Resolved/Verified Bugs

• bug 672642 - make queries static strings in storage
• bug 672655 - Move to a single version number for packaging
• bug 672970 - missing dependencies: libevent-devel, libmemcached-devel
• bug 674768 - Make the default value for auth in baseapp "off"
• bug 671098 - Forces a nullpool when sqlite is used
• bug 675279 - couple of failures detected in Jenkins

Opened Bugs/Issues

• LDAP
  • bug 679063 - need better handling of ldap.SERVER_DOWN
  • bug 679928 - LDAP masters need write cache
  • bug 680048 - make sure scl2 ldap masters are production ready
  • bug 674766 - Put ldapconnector.py into more generic services space
  • bug 674648 - ldapconnection.py add better error response when server is down
  • bug 646945 - Reconfigure LDAP to use uidNumber as the primary key
  • bug 620703 - Figure out a way to create ldap slaves faster.
  • bug 680323 - be more resilient to ldap replication delays
  • bug 680379 - Metabug for all LDAP pool related stuff
  • bug 679063 - need better handling of ldap.SERVER_DOWN
  • bug 674640 - BackendError: BackendError on localhost
  • bug 666536 - LDAPError catch miss on bind
  • bug 668520 - one missing retry on a timeout
  • bug 680324 - ldap connections should have a max lifetime configurable
  • bug 680323 - More considered handling of 401s resulting from Sync LDAP replication delays

• Account Portal
  • bug 679500 - scl2: staging access to external VIPs
  • bug 679501 - support HTTP proxy for deleting user data
  • bug 678557 - server captcha verification needs to support a proxy
  • bug 679960 - account portal build problem with BeautifulSoup pinned to 3.2.0

• Node
  • bug 674280 - if serverURL is broken for a long time, check with node server to be certain it's correct

• Dev/OPs Process
  • bug 677496 - introduce release candidates
  • bug 680360 - Improve the release process

Notes

• This train for Sync Server was actually a combination of three separate deployments:
  • Storage Server (see bug 676109)
  • Account Portal (see bug 670946)
  • sreg (see bug 676351)
• These deployments were delayed beyond 08-10-2011 due to issues in Staging:
  • Delayed configuration due to hardware issues
  • Problems with Sync in Stage: LDAP idle time/server-down (the 503 errors)
• Deployment issues (by order of deployment) also caused issues and a delay.
  • First attempt: Deploy Storage Server, sreg, then Account Portal (this failed)
  • Second attempt: Deploy Storage Server and Account Portal (this succeeded)
  • Third deployment: sreg