(accumulo) 01/01: Merge branch '2.1' into 3.1

[dl...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

dlmarion pushed a commit to branch 3.1
in repository https://gitbox.apache.org/repos/asf/accumulo.git

commit d0fb7f5fd215c9f60dabb123169ef2df71fc6506
Merge: 040a32c640 44b97f36c4
Author: Dave Marion <dl...@apache.org>
AuthorDate: Thu Jan 2 21:21:55 2025 +0000

    Merge branch '2.1' into 3.1

 .../core/fate/zookeeper/DistributedReadWriteLock.java      |  3 ++-
 .../main/java/org/apache/accumulo/server/util/Admin.java   | 14 ++++++++++++--
 .../apache/accumulo/tserver/log/TabletServerLogger.java    | 12 +++++++++---
 3 files changed, 23 insertions(+), 6 deletions(-)

diff --cc server/tserver/src/main/java/org/apache/accumulo/tserver/log/TabletServerLogger.java
index 2aa1e87aa8,46970b4c21..4409602e5e
--- a/server/tserver/src/main/java/org/apache/accumulo/tserver/log/TabletServerLogger.java
+++ b/server/tserver/src/main/java/org/apache/accumulo/tserver/log/TabletServerLogger.java
@@@ -293,66 -325,37 +293,72 @@@ public class TabletServerLogger 
                log.error("Failed to close WAL after it failed to open", e);
              }
  
 -            // it's possible the log was advertised in ZK even though we got an
 -            // exception. If there's a chance the WAL marker may have been created,
 -            // this will ensure it's closed. Either the close will be written and
 -            // the GC will clean it up, or the tserver is about to die due to sesson
 -            // expiration and the GC will also clean it up.
              try {
 -              tserver.walogClosed(alog);
 +              Path path = alog.getPath();
 +              if (fs.exists(path)) {
 +                fs.delete(path);
 +              }
              } catch (Exception e) {
 -              log.error("Failed to close WAL that failed to open: " + fileName, e);
 +              log.warn("Failed to delete a WAL that failed to open", e);
              }
 +          }
  
 -            try {
 -              nextLog.offer(t, 12, TimeUnit.HOURS);
 -            } catch (InterruptedException ex) {
 -              // Throw an Error, not an Exception, so the AccumuloUncaughtExceptionHandler
 -              // will log this then halt the VM.
 -              throw new Error("Next log maker thread interrupted", ex);
 -            }
 +          try {
 +            nextLog.offer(t, 12, TimeUnit.HOURS);
 +          } catch (InterruptedException ex) {
-             // ignore
++            // Throw an Error, not an Exception, so the AccumuloUncaughtExceptionHandler
++            // will log this then halt the VM.
++            throw new Error("Next log maker thread interrupted", ex);
 +          }
 +
 +          continue;
 +        }
  
 -            continue;
 +        log.debug("Created next WAL {}", alog.getLogEntry());
 +
 +        try {
 +          tserver.addNewLogMarker(alog);
 +        } catch (Exception t) {
 +          log.error("Failed to add new WAL marker for " + alog.getLogEntry(), t);
 +
 +          try {
 +            // Intentionally not deleting walog because it may have been advertised in ZK. See
 +            // #949
 +            alog.close();
 +          } catch (Exception e) {
 +            log.error("Failed to close WAL after it failed to open", e);
            }
  
 +          // it's possible the log was advertised in ZK even though we got an
 +          // exception. If there's a chance the WAL marker may have been created,
 +          // this will ensure it's closed. Either the close will be written and
 +          // the GC will clean it up, or the tserver is about to die due to sesson
 +          // expiration and the GC will also clean it up.
            try {
 -            while (!nextLog.offer(alog, 12, TimeUnit.HOURS)) {
 -              log.info("Our WAL was not used for 12 hours: {}", fileName);
 -            }
 -          } catch (InterruptedException e) {
 +            tserver.walogClosed(alog);
 +          } catch (Exception e) {
 +            log.error("Failed to close WAL that failed to open: " + alog.getLogEntry(), e);
 +          }
 +
 +          try {
 +            nextLog.offer(t, 12, TimeUnit.HOURS);
 +          } catch (InterruptedException ex) {
-             // ignore
+             // Throw an Error, not an Exception, so the AccumuloUncaughtExceptionHandler
+             // will log this then halt the VM.
 -            throw new Error("Next log maker thread interrupted", e);
++            throw new Error("Next log maker thread interrupted", ex);
            }
 +
 +          continue;
 +        }
 +
 +        try {
 +          while (!nextLog.offer(alog, 12, TimeUnit.HOURS)) {
 +            log.info("Our WAL was not used for 12 hours: {}", alog.getLogEntry());
 +          }
 +        } catch (InterruptedException e) {
-           // ignore - server is shutting down
++          // Throw an Error, not an Exception, so the AccumuloUncaughtExceptionHandler
++          // will log this then halt the VM.
++          throw new Error("Next log maker thread interrupted", e);
          }
        }
      });