1- # librdkafka v2.11.0
1+ # librdkafka v2.10.1
22
3- librdkafka v2.11.0 is a feature release:
3+ librdkafka v2.10.1 is a maintenance release:
44
5- * Fix for frequent disconnections on push telemetry requests
6- with particular metric configurations (#4912 ).
7- * Avoid copy outside boundaries when reading metric names in telemetry
8- subscription (#5105 )
9- * Metrics aren't duplicated when multiple prefixes match them (#5104 )
5+ * Fix to add locks when updating the metadata cache for the consumer
6+ after no broker connection is available (@marcin-krystianc , #5066 ).
7+ * Fix to the re-bootstrap case when ` bootstrap.servers ` is ` NULL ` and
8+ brokers were added manually through ` rd_kafka_brokers_add ` (#5067 ).
9+ * Fix an issue where the first message to any topic produced via ` producev ` or
10+ ` produceva ` was delivered late (by up to 1 second) (#5032 ).
11+ * Fix for a loop of re-bootstrap sequences in case the client reaches the
12+ ` all brokers down ` state (#5086 ).
13+ * Fix for frequent disconnections on push telemetry requests
14+ with particular metric configurations (#4912 ).
15+ * Avoid copy outside boundaries when reading metric names in telemetry
16+ subscription (#5105 )
17+ * Metrics aren't duplicated when multiple prefixes match them (#5104 )
1018
1119
1220## Fixes
1321
22+ ### General fixes
23+
24+ * Issues: #5088 .
25+ Fix for a loop of re-bootstrap sequences in case the client reaches the
26+ ` all brokers down ` state. The client continues to select the
27+ bootstrap brokers given they have no connection attempt and doesn't
28+ re-connect to the learned ones. In case it happens a broker restart
29+ can break the loop for the clients using the affected version.
30+ Fixed by giving a higher chance to connect to the learned brokers
31+ even if there are new ones that never tried to connect.
32+ Happens since 2.10.0 (#5086 ).
33+ * Issues: #5057 .
34+ Fix to the re-bootstrap case when ` bootstrap.servers ` is ` NULL ` and
35+ brokers were added manually through ` rd_kafka_brokers_add ` .
36+ Avoids a segmentation fault in this case.
37+ Happens since 2.10.0 (#5067 ).
38+
39+ ### Producer fixes
40+
41+ * In case of ` producev ` or ` produceva ` , the producer did not enqueue a leader
42+ query metadata request immediately, and rather, waited for the 1 second
43+ timer to kick in. This could cause delays in the sending of the first message
44+ by up to 1 second.
45+ Happens since 1.x (#5032 ).
46+
47+ ### Consumer fixes
48+
49+ * Issues: #5051 .
50+ Fix to add locks when updating the metadata cache for the consumer.
51+ It can cause memory corruption or use-after-free in case
52+ there's no broker connection and the consumer
53+ group metadata needs to be updated.
54+ Happens since 2.10.0 (#5066 ).
55+
1456### Telemetry fixes
1557
16- * Issues: #5106
58+ * Issues: #5106 .
1759 Fix for frequent disconnections on push telemetry requests
1860 with particular metric configurations.
1961 A ` NULL ` payload is sent in a push telemetry request when
@@ -22,65 +64,17 @@ librdkafka v2.11.0 is a feature release:
2264 some metrics are matching the producer but none the consumer
2365 or the other way around.
2466 Happens since 2.5.0 (#4912 ).
25- * Issues: #5102
67+ * Issues: #5102 .
2668 Avoid copy outside boundaries when reading metric names in telemetry
2769 subscription. It can cause that some metrics aren't matched.
2870 Happens since 2.5.0 (#5105 ).
29- * Issues: #5103
30- Metrics aren't duplicated when multiple prefixes match them.
71+ * Issues: #5103 .
72+ Telemetry metrics aren't duplicated when multiple prefixes match them.
3173 Fixed by keeping track of the metrics that already matched.
3274 Happens since 2.5.0 (#5104 ).
3375
3476
3577
36- # librdkafka v2.10.1
37-
38- librdkafka v2.10.1 is a maintenance release:
39-
40- * Fix to add locks when updating the metadata cache for the consumer
41- after no broker connection is available (@marcin-krystianc , #5066 ).
42- * Fix to the re-bootstrap case when ` boostrap.servers ` is ` NULL ` and
43- brokers were added manually through ` rd_kafka_brokers_add ` (#5067 ).
44- * Fix an issue where the first message to any topic produced via ` producev ` or
45- ` produceva ` was not delivered late (by up to 1 second) (#5032 ).
46- * Fix for a loop of re-bootstrap sequences in case the client reaches the
47- ` all brokers down ` state (#5086 ).
48-
49-
50- ## Fixes
51-
52- ### Producer fixes
53-
54- * In case of ` producev ` or ` produceva ` , the producer did not enqueue a leader
55- query metadata request immediately, and rather, waited for the 1 second
56- timer to kick in. This could cause delays in the sending of the first message
57- by up to 1 second. (#5032 ).
58-
59- ### Consumer fixes
60-
61- * Issues: #5051
62- Fix to add locks when updating the metadata cache for the consumer.
63- It can cause memory corruption or use-after-free in case
64- there's no broker connection and the consumer
65- group metadata needs to be updated.
66- Happens since 2.10.0 (#5066 ).
67- * Issues: #5057
68- Fix to the re-bootstrap case when ` boostrap.servers ` is ` NULL ` and
69- brokers were added manually through ` rd_kafka_brokers_add ` .
70- Avoids a segmentation fault in this case.
71- Happens since 2.10.0 (#5067 ).
72- * Issues: #5088
73- Fix for a loop of re-bootstrap sequences in case the client reaches the
74- ` all brokers down ` state. The client continues to select the
75- bootstrap brokers given they have no connection attempt and doesn't
76- re-connect to the learned ones. In case it happens a broker restart
77- can break the loop for the clients using the affected version.
78- Fixed by giving a higher chance to connect to the learned brokers
79- even if there are new ones that never tried to connect.
80- Happens since 2.10.0 (#5086 ).
81-
82-
83-
8478# librdkafka v2.10.0
8579
8680librdkafka v2.10.0 is a feature release:
0 commit comments