Back to project
Issues and filters
View all filters
"Jonathan's Test Case" causes master to crash in bfac47df7c
RamCloud client object has memory leaks
Provide user a way to quickly get stats on each master server
(Multiple?) Context problem involving dropped pollers and wedged RPCs
stderr is closed in some unit test
Deadlock during TransactionManager destruction
Better documentation for Segment::syncedLength
RAMCloud reports Logger got stuck when executing clusterperf.py writeThroughput
Solve ABA problem for secondary index without holding HashTable bucket lock.
Delete overflow buckets in HashTable
Handle server crashes during split/migrate of indexlet
Lots of data race reports generated by ThreadSanitizer when running unit tests
Remove the deprecated synchronous calls to ObjectFinder
clientTests/RAM-462/ram462test is broken
Btree dependencies not handled correctly?
Assertion failure in Segment::EntryHeader::getType
Transactions still deadlocking over WorkerTimer
Mechanism to detect "TX timeout is too short"
Give error if a client tries to write data > 1MB?
Double check tableStat correctness for linearizability & tx
Recovery.py aborts with tcp transport.
Migrate from JIRA to GitHub issues
Avoid 64-bit modulo in UnackedRpcResult
RamCloud object constructor comments falsely claim it throws CouldntConnectException
SafeVersion log entry not migrated
Drop table hashtable cleanup underflows Log's totalLiveBytes
Instability if InfRc transmit buffers exhausted
Implement fault tolerance for coordinator functions related to indexing
Change references "key" => "primaryKey"
In splitAndMigrateIndexlet, need to keep track of ctime
For splitAndMigrateIndexlet, add missing unit tests
More efficient locking in B+ Tree
IndexBtree segfaults while emplaceAppend-ing an InnerNode to a Buffer and IndexBtree::innerslotmax = 32 && DEBUG=NO
CoordinatorServerListTest.updateLoopAndWaitForWork seems to be timing dependent.
Support secondary indexing features in transaction.
Use RAMCloud tables to store lease information.
IndexBtree's bounds definition is inconsistent with the rest of RAMCloud
TableManager::coordSplitAndMigrateIndexlet holds a lock for the entire duration of the operation
Batch mode for CoordinatorClient::getLeaseInfo
Test failure in linearizability branch
The lost-open-replica protocol doesn't get induced if an open replica has already experienced a failure.
If index entries cannot be appended to the log, then this needs to be propagated back to the caller.
If master runs out of space during recovery, it should abort the recovery.
While replaying a segment, handle the case where the entry has an invalid checksum properly.
recovery.py fails with smaller number of object (eg. recovery.py -n 296475)
In InfRcTransport::clientTryExhangeQueuePairs, there is a safety violation.
HashTable destructor should free chained CacheLines that were allocated in insert()
Add unit tests for notifyCreateIndex and notifyDropIndex in TableManagerTest
Add unit tests for reference_getEntry_contiguous and reference_getEntry_discontiguous to SegmentTest
issue 1 of 244