[TEZ-3812] race condition in ssl shuffle

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Assignee: Unassigned
• Reporter: Sergey Shelukhin
• Created: 03/Aug/17 19:31
• Updated: 03/Aug/17 19:32

Description

ShuffleUtils does the following:

if (sslFactory == null) {
        synchronized (HttpConnectionParams.class) {
          //Create sslFactory if it is null or if it was destroyed earlier
          if (sslFactory == null || sslFactory.getKeystoresFactory().getTrustManagers() == null) {
            sslFactory =
                new SSLFactory(org.apache.hadoop.security.ssl.SSLFactory.Mode.CLIENT, conf);
            try {
              sslFactory.init();

It is possible for a thread to get sslFactory that has been assigned but not initialized. It could result in e.g. the hostnameVerifier being null:

Caused by: java.lang.IllegalArgumentException: no HostnameVerifier specified
at javax.net.ssl.HttpsURLConnection.setHostnameVerifier(HttpsURLConnection.java:265)
at org.apache.tez.http.SSLFactory.configure(SSLFactory.java:219)
at org.apache.tez.http.HttpConnection.setupConnection(HttpConnection.java:98)
at org.apache.tez.http.HttpConnection.connect(HttpConnection.java:137)
at org.apache.tez.http.HttpConnection.connect(HttpConnection.java:123)
at org.apache.tez.runtime.library.common.shuffle.orderedgrouped.FetcherOrderedGrouped.setupConnection(FetcherOrderedGrouped.java:340)
at org.apache.tez.runtime.library.common.shuffle.orderedgrouped.FetcherOrderedGrouped.copyFromHost(FetcherOrderedGrouped.java:260)
at org.apache.tez.runtime.library.common.shuffle.orderedgrouped.FetcherOrderedGrouped.fetchNext(FetcherOrderedGrouped.java:178)
at org.apache.tez.runtime.library.common.shuffle.orderedgrouped.FetcherOrderedGrouped.callInternal(FetcherOrderedGrouped.java:191)
at org.apache.tez.runtime.library.common.shuffle.orderedgrouped.FetcherOrderedGrouped.callInternal(FetcherOrderedGrouped.java:54)