2019-11-07 09:22:52 UTC - Zhiyong Li: Hello team, how can i set parameters such as web-service-url & broker-service-url while initializing cluster metadata without a DNS name? I think I can use multi-host url for broker-service-url, but how about web-service-url ? Could you please give me an example? Thanks. ---- 2019-11-07 09:38:03 UTC - Sijie Guo: yes.
<pulsar://host1:6650,host2:6650> <http://host1:8080,host2:8080> ---- 2019-11-07 09:42:54 UTC - Alexandre DUVAL: + you agree that fill an header http to a ws connection is not ws RFC compatible? ---- 2019-11-07 10:33:20 UTC - hangc: @hangc has joined the channel ---- 2019-11-07 10:51:50 UTC - listar: @listar has joined the channel ---- 2019-11-07 10:53:16 UTC - leozhao: @leozhao has joined the channel ---- 2019-11-07 14:09:21 UTC - Alexandre DUVAL: another question, when a pulsar function write to multiple topic using context.newoutputmessage, the return fonction type is void. How can I "make" the function process fail and nack the message for some reasons? ---- 2019-11-07 14:13:44 UTC - Alexandre DUVAL: @Jerry Peng hi, which yaml field can be used to define userConfigMap can be used in function body? ---- 2019-11-07 14:17:24 UTC - Alexandre DUVAL: just throw an uncatch exception? ---- 2019-11-07 14:29:57 UTC - Benjamin Egelund-Müller: @Benjamin Egelund-Müller has joined the channel ---- 2019-11-07 15:14:52 UTC - Sijie Guo: yeah that works ---- 2019-11-07 15:55:09 UTC - Ciaran McManus: @Ciaran McManus has joined the channel ---- 2019-11-07 17:23:20 UTC - Igor Zubchenok: Hi! We updated to 2.4.1 pulsar and have a lot of these errors: >ERROR org.apache.bookkeeper.proto.checksum.CRC32CDigestManager - Sse42Crc32C is not supported, will use a slower CRC32C implementation. How to fix this? ---- 2019-11-07 18:48:15 UTC - Luke Lu: RFC 6455 section 4.1 explicitly allows that: “12. The request MAY include any other header fields, for example cookies [RFC6265] and/or authentication-related header fields such as the |Authorization| header field [RFC2616], which are processed according to documents that define them.” ---- 2019-11-07 19:33:07 UTC - Joshua Dunham: Hi Everyone, does anyone have a config for TLS termination using something like HAProxy then forwarding TCP to Pulsar? ---- 2019-11-07 19:36:19 UTC - Matteo Merli: I don't have HAProxy conf, though you can terminate the TLS on the Pulsar proxy ---- 2019-11-07 19:40:45 UTC - Joshua Dunham: Any client side issues to worry about or is it (python) generally oblivious? ---- 2019-11-07 19:42:37 UTC - Matteo Merli: There shouldn't be any surprises ---- 2019-11-07 19:49:44 UTC - Joshua Dunham: Nice! Thx ---- 2019-11-07 19:55:32 UTC - Vlad Lazarenko: Hello Everyone. Running into a very strange performance issue. Expiry and TTL on a topic is set to 3 days. There are only about 8000 messages. If topic wasn’t touched for about a day and I start a consumer with cursor set to earliest, the replay is painfully slow - it takes 7 seconds. The throughput tested with pulsar-perf is anywhere from 497 to 1970 msg/s. Subsequent replay takes a second with 8055 msg/sec. Running on bare metal, almost default settings, hosts are not loaded at all... what could be an issue? ---- 2019-11-07 20:05:38 UTC - Vlad Lazarenko: Also, the messages are small. So it is 0.7 Mbit/s vs 3.05 for subsequent runs. Nowhere near any HW limits. ---- 2019-11-07 20:23:37 UTC - Addison Higham: what sort of disks in bookkeeper? ---- 2019-11-07 21:43:28 UTC - Oleg Kozlov: Hello, we are setting up Pulsar on Kubernetes, and getting an issue similar to <https://github.com/apache/pulsar/issues/5206>, however in our case we run Pulsar on Kubernetes and trying to have SSL terminate on the load balancer setup in front of the Pulsar Proxy, so the client has to send msgs using "pulsar+ssl", but it's would be decrypted behind the load balancer, so it's not clear whether proxy and broker should use standard broker port 6650, or TLS broker port 6651? Can someone take a look and help us resolve? ---- 2019-11-07 22:04:14 UTC - Oleg Kozlov: At this point we are mostly interested if we are doing the right thing with setup, enabling TLS on the client, but not on the Proxy & Broker, since our theory is that the kubernetes load balancer takes care of decryption ---- 2019-11-07 22:06:55 UTC - Vlad Lazarenko: SSD. And special tweaks for that? That was so slow that I can’t really blame a disc, I think even a tape should produce faster results ---- 2019-11-07 22:19:39 UTC - Rafal Gorzkowski: @Rafal Gorzkowski has joined the channel ---- 2019-11-07 22:58:39 UTC - Britt Bolen: Actually looking at the zts-client-java docs it seems that I should be able to pass the path and password of a trust store through System.setProperty() but haven’t seen it work yet… will need to dig deeper into the athenz/pulsar code… ---- 2019-11-07 22:59:44 UTC - Britt Bolen: ---- 2019-11-08 01:22:05 UTC - Igor Zubchenok: up! ---- 2019-11-08 02:04:39 UTC - Sijie Guo: what type of os are you running this? ---- 2019-11-08 02:11:20 UTC - Sijie Guo: the error indicates that your hardware doesn’t support sse42, so it fallls back to use a java implementation. ---- 2019-11-08 02:12:16 UTC - Sijie Guo: We changed the default checksum to crc32c. it used to be crc32. ---- 2019-11-08 02:12:24 UTC - Igor Zubchenok: Ubuntu 18.04.3 LTS ```Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 8 On-line CPU(s) list: 0-7 Thread(s) per core: 2 Core(s) per socket: 4 Socket(s): 1 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 158 Model name: Intel(R) Core(TM) i7-7700 CPU @ 3.60GHz Stepping: 9 CPU MHz: 4000.000 CPU max MHz: 4200.0000 CPU min MHz: 800.0000 BogoMIPS: 7200.00 Virtualization: VT-x L1d cache: 32K L1i cache: 32K L2 cache: 256K L3 cache: 8192K NUMA node0 CPU(s): 0-7 Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf tsc_known_freq pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb invpcid_single pti ssbd ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm mpx rdseed adx smap clflushopt intel_pt xsaveopt xsavec xgetbv1 xsaves dtherm ida arat pln pts hwp hwp_notify hwp_act_window hwp_epp md_clear flush_l1d``` ---- 2019-11-08 02:12:55 UTC - Sijie Guo: the message should be warn. there is already a fix in bookkeeper. so it is harmless. ---- 2019-11-08 02:13:47 UTC - Igor Zubchenok: Yes, it is harmless except massive logging on WARN level. However what is the issue since CPU supports sse4_2? ---- 2019-11-08 02:17:08 UTC - Sijie Guo: you can switch to crc32 to avoid the WARN level. configure the following setting in your broker.conf `managedLedgerDigestType=CRC32` ---- 2019-11-08 02:18:08 UTC - Igor Zubchenok: Ok, I see. BTW I much more interested to use crc32c. ---- 2019-11-08 02:38:48 UTC - Matteo Merli: seems the JNI might not be compiled for linux ---- 2019-11-08 02:39:03 UTC - Matteo Merli: that's why it falls back to Java based implementation ---- 2019-11-08 02:58:00 UTC - Sijie Guo: @Matteo Merli I checked 4.9.2. it seems to be fine. it is a linux lib. ---- 2019-11-08 02:58:25 UTC - Sijie Guo: since you were releasing 4.9.2, did you remember how do you compile the artifacts? ---- 2019-11-08 03:30:51 UTC - Igor Zubchenok: How can I add this JNI lib to existing build of 2.4.1 of pulsar? ----
