Skip to content
Fetching contributors…
Cannot retrieve contributors at this time
113 lines (107 sloc) 5.77 KB
arakoon (0.8.2-1) unstable; urgency=low
This cycle the focus was on bugfixing.
Full details are in the Arakoon Jira. Here's the quick list
* [ARAKOON-133] Arakoon daemon takes GBytes of memory when doing large catchup
in combination with debug logging enabled
* [ARAKOON-132] Arakoon server crashed as a result of
'invalid store update request'
* [ARAKOON-131] Catchup fails if it starts from a multiple of the number
of tlog entries per tlog file
* [ARAKOON-130] Slave cannot promise on same n more than once
* [ARAKOON-129] Arakoon node runs out of file descriptors when compressing
* [ARAKOON-128] Slave considers master lease expired when it's not
* [ARAKOON-127] Log last n debug log messages when the state machine crashes
* [ARAKOON-126] There are still some corner cases where we probably don't do
the right thing
* [ARAKOON-125] In case of master reelection or catchup slaves do not always
push previous uncommited updates in the store
* [ARAKOON-124] Masterlease uses absolute timestamps as part of the update
* [ARAKOON-123] Log rotation should reload log configuration
* [ARAKOON-122] expectprogress returns incorrect or no value in forced master
setup when slave is not running
* [ARAKOON-121] Slave dies when receiving promis which n is higher than
* [ARAKOON-120] Restarting a forced master at the wrong time can lead to
cluster without progress
* [ARAKOON-119] Node should die when a corrupt tc is encountered at runtime
* [ARAKOON-118] tlog iteration starting from the wrong value starts
in the wrong tlog file
* [ARAKOON-117] When a slave goes to election mode, the previous update is
not always pushed to the store
* [ARAKOON-115] Updates can be pushed into the store twice which under
certain conditions can lead to data loss
* [ARAKOON-114] Interuption of tlog compression results in half compressed
tlogs which are treated as a normal tlog
* [ARAKOON-112] when arakoon node is down, but machine is up, connecting
nodes hammer
* [ARAKOON-111] Monkey dies with "Lwt.wakeup invalid argument"
* [ARAKOON-110] State machine fails when running as master, it receives
a message from newly elected master
* [ARAKOON-109] Election timeout is not always fired with can lead
to master-less cluster
* [ARAKOON-107] Specifying tlog directory with Pylabs Arakoon extension
results in error
* [ARAKOON-106] Master election can get in a loop
* [ARAKOON-105] If system clock is changed lease is not invalidated
* [ARAKOON-103] Number of tlogs is limited to 1000
* [ARAKOON-102] Provide clone functionality
* [ARAKOON-101] Improve catchup speed
* [ARAKOON-100] Provide basic load statistics
* [ARAKOON-99] Store can get ahead of tlogs
* [ARAKOON-98] Empty node can become master
* [ARAKOON-97] Automatically truncate tlogs if they are corrupt
* [ARAKOON-95] Add repair option that can be used in case of tlog corruption
* [ARAKOON-94] having a machine that is not live in an arakoon config makes
the cluster slow
* [ARAKOON-93] Master restart can lead to a missed update in the slave store
* [ARAKOON-92] Retry during master election does not always retry correctly
* [ARAKOON-91] Arakoon getStatus() and getStatusOne returns unwanted
PID information
* [ARAKOON-90] node dies if too many clients connect
* [ARAKOON-85] Failing system test: Disk full on slave
* [ARAKOON-83] Add possibility to define clusterid for a certain set of
Arakoon nodes
* [ARAKOON-82] Add protocol version check
* [ARAKOON-81] With forced master, the master's store counter can get
ahead of the tlogs
* [ARAKOON-79] Tlog rollover can make catchup fail
* [ARAKOON-78] expect_progress_possible returns true in scenarios where
a slave is in catchup
* [ARAKOON-77] Lease can expire without being stored
* [ARAKOON-76] multicast of messages is serial, which sometimes gives
bad performance
* [ARAKOON-71] Run Arakoon as a different user
arakoon (0.8.1-1) unstable; urgency=low
Full details are in the Arakoon JIRA. Here's the quick list
* [ARAKOON-67] Provide TC tools in QPackage
* [ARAKOON-76] Multicast messages is serial which sometimes gives
very bad performance
* [ARAKOON-75] Master Election issue in degraded state after restarting
* [ARAKOON-61] Frequently restarting the master node in a cluster node
under load scenario might lead to cluster unavailability
* [ARAKOON-64] multi-get support
* [ARAKOON-48] Request the state of a server node
* [ARAKOON-66] Delete of non-existing key messes up tlog and store counters
* [ARAKOON-63] Transactions lead to non-transactional behaviour
* [ARAKOON-62] if one operation failse because the key does not exist,
sequence does not return that key
* [ARAKOON-52] Userot test_and_set can't see the current value was his
new value
* [ARAKOON-43] Report name of key when raising error: 'Key not found'
* [ARAKOON-58] Config file cannot contain 'empty' lines with spaces
* [ARAKOON-57] Last entry goes missing after master goes down
* [ARAKOON-56] q.config.arakoon.tearDown() does not remove config file entry
for fixed master
* [ARAKOON-55] Missing entry in tlog after master changeO
* [ARAKOON-54] An empty node can become master just after it starts
* [ARAKOON-53] ArakoonExceptions.ArakoonNodeNotMaster:
Cannot perform operation on non-master node
* [ARAKOON-46] Keys is present in tlog but not in tc on some nodes
of the cluster
* [ARAKOON-44] Arakoon deamon should die if it detects a corrupt tlog
arakoon (0.8.0-1) unstable; urgency=low
* Initial Public Release. Please bear with us
-- Romain Slootmaekers <> Mon, 25 Oct 2010 21:40:27 +0200
Jump to Line
Something went wrong with that request. Please try again.